Slow motion replay real-time frame insertion control system based on push rod operating state

Abstract

The present application relates to the technical field of image communication, and especially provides a slow motion playback real-time frame insertion control system based on a push rod operation state. The system comprises a push rod state acquisition and zero calibration module, an operation state determination and state switching module, a playback time axis target generation module, a time domain reconstruction budget allocation module, a double-channel strategy decision module, a double-stage frame insertion execution module, a versioned buffer and seamless replacement output module, and a constraint closed loop and hierarchical load reduction module. The system guarantees low delay time axis followability, improves the display quality of local slow motion segments in the stable residence stage, and enhances the controllability and stability of the system under actual operation constraints.

Description

Technical Field

[0001] This invention relates to the field of image communication technology, and in particular to a real-time frame interpolation control system for slow-motion playback based on the state of a push rod operation. Background Technology

[0002] In scenarios involving precise video playback, such as sports event replay, film and television post-production review, and broadcast control, the playback direction and speed are often controlled using sliders, joysticks, or other continuous position input devices. During slow-motion playback, to improve display smoothness, intermediate frames are typically generated between the original video frames. Existing solutions generally determine the number of interpolated frames and processing intensity uniformly based on the playback magnification, input / output frame rate, or device load. However, the slider control process usually has distinct stages: when the slider moves rapidly, the system requires low-latency following; after the slider decelerates and stabilizes, the operator focuses more on the viewing quality of specific slow-motion segments. Existing technologies often process the entire playback process using the same interpolation mode, or primarily determine the number and intensity of interpolated frames based on the current playback magnification, lacking a mechanism for layered control based on differences in the operational stages. Therefore, it is difficult to simultaneously meet the high-quality display requirements of both the rapid following and stable stabilization stages.

[0003] In addition, existing frame interpolation results are usually buffered and output sequentially as a single frame to be displayed, lacking a multi-version management structure for the same display moment. This makes it difficult for the high-quality results generated later to replace the target moment that has not yet been displayed without disrupting the current playback timeline. At the same time, the resource control of the existing system also lacks a closed-loop control mechanism that links with factors such as the display deadline, whether the high-quality results can actually take effect, and whether the current operation status is stable. Summary of the Invention

[0004] In view of this, the present invention provides a real-time frame interpolation control system for slow-motion playback based on the state of push rod operation, which improves the display quality of local slow-motion segments during the stable dwell phase while ensuring low-latency time axis tracking, and enhances the controllability and stability of the system under actual operating constraints.

[0005] In a first aspect, the present invention provides a real-time frame interpolation control system for slow-motion playback based on the state of a push rod operation, the system comprising:

[0006] The module includes: push rod status acquisition and zero-position calibration module, operation status determination and status switching module, playback time axis target generation module, time domain reconstruction budget allocation module, dual-path strategy decision module, dual-stage frame interpolation execution module, versioned buffer and seamless replacement output module, and constraint closed loop and graded load reduction module. The push rod status acquisition and zero-position calibration module is used to continuously acquire the position sequence of the push rod, perform dead zone calibration on the small jitters near the zero position of the push rod, and output the standardized position value P(t), zero-crossing event and dwell time information. The operation state determination and state switching module is used to extract displacement change, velocity, acceleration or velocity change trend and dwell time based on the position sequence, zero crossing event and dwell time information, and divide the current push rod operation into rapid crossing phase, approach positioning phase and stable dwell phase according to the preset state rules, and output the current control state S_ctrl. The playback timeline target generation module generates target playback timeline parameters based on the standardized position value P(t) and the current control state S_ctrl. When the current control state S_ctrl is in a stable dwell phase and the target playback timeline parameters remain stable within a number of consecutive sampling periods, a backfill trigger flag is output. The target playback timeline parameters include the playback direction, time scaling ratio, and the advance rate of the current playback position on the original video timeline. The temporal reconstruction budget allocation module is used to generate a preview budget B_pre, a backfill budget B_ref, and a local backfill window W_ref based on the current control state S_ctrl, the target playback timeline parameters, the display time limit constraints, and the system running state. The preview budget B_pre is used to constrain the resource investment in generating candidate intermediate frames in the preview path, the backfill budget B_ref is used to constrain the resource investment in generating alternative frame versions in the backfill path, and the local backfill window W_ref is used to limit the time range or frame range that allows high-quality backfill to be performed. The dual-path strategy decision module is used to determine the current frame interpolation path organization method that the system should adopt among the preview path, backfill path, and safety path based on the current control state S_ctrl, the target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref, and to perform a display cutoff reachability judgment on candidate intermediate frames or alternative frame versions. Specifically, the preview path is used to prioritize generating candidate intermediate frames that meet the current display cutoff time requirement to ensure timeline continuity and low-latency response; the backfill path is used to generate high-quality alternative frame versions for candidate intermediate frames within the local backfill window that have not yet reached the display cutoff time during the stable dwell phase to improve local slow-motion image quality; the safety path is used to maintain continuous output of the current display timeline in a low-load manner when the system processing capacity is insufficient, the display cutoff risk is increased, or high-quality results cannot take effect on time. The two-stage frame interpolation execution module is used to perform low-latency frame interpolation on the pair of frames to be output based on the preview budget B_pre, generate candidate intermediate frames that meet the current display deadline requirements, and bind the corresponding display timestamp to the candidate intermediate frames; and when the backfill trigger flag is valid and the current control state S_ctrl is in a stable dwell phase, it performs high-quality frame interpolation on the candidate intermediate frames in the local backfill window W_ref that have not yet reached their corresponding display deadline, so as to generate a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; The versioned buffer and seamless replacement output module is used to establish buffer entries according to the display timestamp, perform versioned management of the candidate intermediate frame and the replacement frame version, and when the replacement frame version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the current valid output version is updated from the candidate intermediate frame to the replacement frame version, and the current valid output version is output according to the display timestamp order. The constraint closed-loop and graded load reduction module is used to monitor the display cutoff default risk, path processing load, buffer filling status, task queue backlog, backfill efficiency, and current control state stability. Based on the monitoring results, it performs graded load reduction and load recovery to prioritize the continuous output and display timeliness of the current playback timeline.

[0007] Optionally, the fader state acquisition and zero-position calibration module performs dead-zone calibration on the tiny jitters near the zero position of the fader and outputs standardized position values, zero-crossing events, and dwell time information; the zero-crossing events are used for subsequent stable switching of playback direction, and the dwell time information is used to determine whether the system has entered a stable dwell phase, thereby providing a trigger basis for two-stage frame interpolation and local high-quality backfill; To avoid state jitter near critical conditions, the operation state determination and state switching module sets entry and exit thresholds between adjacent states to form a hysteresis interval. The operation state determination and state switching module outputs the current control state S_ctrl and its switching event, providing a basis for subsequent playback timeline target generation, budget allocation and path decision-making. The playback timeline target generation module sets a freeze zone near the zero position to keep the system stable when switching between normal playback and slow motion playback. When the operation state enters the stable dwell phase, the preset dwell conditions are met, and the target playback timeline parameters remain stable within several consecutive sampling periods, the playback timeline target generation module outputs a high-quality backfill trigger flag to notify subsequent modules to perform higher-quality backfill processing on the frames to be displayed that have not yet reached the display deadline. In the fast-crossing phase, the temporal reconstruction budget allocation module prioritizes allocating the preview budget to meet the low-latency output requirements; in the approach positioning phase, the temporal reconstruction budget allocation module gradually establishes the backfill budget and backfill window; in the stable dwell phase, the temporal reconstruction budget allocation module increases the backfill budget and determines the local backfill window while ensuring that the preview path outputs on time. The dual-path strategy decision module performs an explicit deadline reachability judgment on the current path organization method. This is used to determine whether the candidate version can still be generated and used for output before the explicit deadline of the corresponding explicit timestamp, and whether the alternative version can still be generated before the explicit deadline of the corresponding explicit timestamp and complete the replacement before the corresponding buffer entry enters the output locking boundary and thus take effect. When the judgment result shows that the current path cannot meet the explicit constraints, the backup path is triggered to intervene. When the dual-stage frame interpolation execution module enters a stable dwell phase and continues for more than a preset duration, it initiates background partial backfilling for a preset time window in the vicinity of the current playback position. The partial backfilling re-executes high-quality frame interpolation for candidate intermediate frames within the time window that have not yet reached the display deadline, so as to obtain a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame. Through the dual-stage execution method of previewing first and then backfilling, the system prioritizes ensuring follow-up when the slider moves quickly and prioritizes improving local image quality after the slider stabilizes. The versioned buffer and seamless replacement output module maintains at least two frame versions for the same display time: a candidate version generated by the preview path and a replacement version generated by the backfill path. Each version shares the same display timestamp but has different version identifiers, validity statuses, and priority attributes. The versioned buffer and seamless replacement output module establishes a versioned buffer structure indexed by the display timestamp. When the replacement version corresponding to the current display timestamp is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the valid output version corresponding to the current display timestamp is updated from the candidate version to the replacement version, while maintaining its display timestamp. When the replacement version fails to be generated before the deadline, or the corresponding buffer entry has entered the output locking boundary, the candidate version is maintained as the valid output version for the current display timestamp. During output control, the system selects the highest priority version currently available for display based on the display timestamp and version validity flag. When the system is in a fast pass-through phase, the constraint-loop and graded load reduction module prioritizes reducing the backfill budget, compressing the backfill window, or temporarily disabling the backfill path to ensure timeline tracking and low-latency output. When the system is in a stable dwell phase, the constraint-loop and graded load reduction module prioritizes shortening the local backfill window, reducing the backfill complexity level, or reducing the number of local frames to be rebuilt. When necessary, the constraint-loop and graded load reduction module switches to the safety path strategy to avoid buffer underflow or continuous accumulation of display defaults. When the system load recovers and continues to meet the preset conditions, the constraint-loop and graded load reduction module gradually restores the backfill budget and local backfill window length in the reverse order of load reduction.

[0008] Optionally, the push rod state acquisition and zero-position calibration module includes a position acquisition and standardization submodule and a zero-position dead zone calibration and event output submodule; The position acquisition and standardization submodule continuously acquires the displacement information of the push rod and converts the acquired raw displacement into a standardized position value that can be used for subsequent control. The submodule obtains the raw position sample values ​​of the push rod and normalizes them according to the push rod's mechanical stroke range to obtain the standardized position value P(t). The submodule outputs a position sequence arranged chronologically and adds sampling timing information to each sample value for subsequent modules to calculate dynamic characteristics such as displacement change, velocity, acceleration, and dwell time. The submodule performs basic stabilization processing on the raw sample values ​​to suppress the impact of instantaneous spike noise or abnormal jumps on subsequent control. The final output of the position acquisition and standardization submodule includes at least: the raw position value, the standardized position value P(t), and the corresponding sampling timing information. The zero-position dead zone calibration and event output submodule is used to perform dead zone calibration on minute jitters near the zero position of the push rod, and outputs control event information related to the zero position based on the standardized position sequence. The zero-position dead zone calibration and event output submodule establishes a zero-position dead zone around the push rod zero position, the boundary of which is determined based on a preset threshold, equipment calibration results, or historical fluctuation statistics. The zero-position dead zone is set using a fixed threshold method or corrected using an adaptive method based on historical fluctuation statistics. When the standardized position value P(t) falls within the zero-position dead zone range, the zero-position dead zone calibration and event output submodule outputs a zero-position locked state. When the standardized position value P(t) moves from one side of the zero-position dead zone... When crossing to the other side, the zero-position dead zone calibration and event output submodule outputs a zero-crossing event signal; when the standardized position value P(t) remains in a low-change state near the zero position or the current position and continues for more than a preset duration, the zero-position dead zone calibration and event output submodule accumulates dwell time information; to avoid frequent entry and exit jitter near the zero-position boundary, the zero-position dead zone calibration and event output submodule sets an entry threshold and an exit threshold at the zero-position dead zone boundary, so that the establishment and release of the zero-position locking state have hysteresis characteristics; the final output of the zero-position dead zone calibration and event output submodule includes at least: zero-position locking state, zero-crossing event, dwell time value, and timing information corresponding to the event.

[0009] Optionally, the operation state determination and state switching module includes a dynamic feature extraction submodule, a state determination submodule, and a hysteresis switching and state output submodule; The dynamic feature extraction submodule receives standardized position values ​​P(t), zero-position locking status, zero-crossing events, and dwell time information, and extracts dynamic features for operation state determination based on continuous position sequences. The submodule maintains a position sequence or sliding time window updated chronologically, and calculates at least one or more of the following dynamic features: displacement change, velocity, acceleration or velocity change trend, dwell time, position fluctuation amplitude, and zero-position related state information. To improve the stability of dynamic features, the submodule smooths the velocity and acceleration to suppress interference from mechanical micro-vibrations, sampling noise, or short-term abnormal jumps. The submodule outputs a set of feature parameters for state determination, which are then used for subsequent state determination and state switching.

[0010] The state determination submodule is used to determine the current push rod operation state based on the feature parameters output by the dynamic feature extraction submodule, and classify it into one of the following: rapid crossing phase, approach positioning phase, and stable dwell phase. When the absolute value of the push rod's movement speed is higher than a preset first speed threshold and the position change is continuously large within the current time window, it is determined to be a rapid crossing phase. When the absolute value of the push rod's movement speed is less than the first speed threshold but greater than a preset second speed threshold, and the speed change trend shows a decreasing trend, it is determined to be an approach positioning phase. When the position change of the push rod within the current time window is less than a preset position change threshold and the duration is greater than a preset dwell time threshold, it is determined to be a stable dwell phase. The first speed threshold, the second speed threshold, the preset position change threshold, and the preset dwell time threshold are set according to the equipment calibration results, the push rod's mechanical stroke, the position sampling resolution, or the historical operation statistics results.

[0011] The hysteresis switching and state output submodule is used to stabilize the state determination result and output the final effective control state. The hysteresis switching and state output submodule sets entry and exit thresholds between adjacent states to give the state switching hysteresis characteristics, avoiding frequent state jumps due to short-term fluctuations near the threshold boundaries. The hysteresis switching and state output submodule combines zero-position locking state and zero-crossing events to constrain state switching. When the system has entered a stable dwell phase, if a new zero-crossing event, continuous large position changes, or a significant re-acceleration trend is detected, it exits the stable dwell phase and switches to the approach positioning phase or rapid crossing phase. The hysteresis switching and state output submodule ultimately outputs the current control state S_ctrl and the corresponding state switching event information for subsequent playback timeline target generation, budget allocation, path selection, and load reduction recovery control.

[0012] Optionally, the playback timeline target generation module includes a timeline parameter generation submodule and a backfill trigger flag generation submodule; The timeline parameter generation submodule receives the standardized position value P(t), zero-position locking status, zero-crossing event, and current control state S_ctrl, and generates target playback timeline parameters based on the above inputs. The target playback timeline parameters include at least the playback direction, time scaling ratio, and the current playback position's advance rate on the original video timeline. The timeline parameter generation submodule generates the playback direction and time scaling ratio based on the magnitude and sign of the standardized position value P(t), and sets a freeze zone near the zero position. When the standardized position value P(t) falls within the freeze zone, the timeline parameter generation submodule maintains the current valid timeline parameters or smoothly transitions to normal playback according to preset rules to avoid frequent changes in playback direction or advance rate due to slight oscillations of the push bar near the zero position. Upper and lower limits are set for the time scaling ratio and playback position advance rate to limit abrupt changes or excessively slow advance of the timeline caused by abnormal inputs. The timeline parameter generation submodule outputs the current playback direction, time scaling ratio, playback position advance rate, and corresponding timing information. The backfill trigger flag generation submodule combines the current control state S_ctrl with the target playback time axis parameters to determine whether the system meets the basic conditions for entering the high-quality local backfill stage, and outputs a backfill trigger flag when the conditions are met. When the system is in a stable dwell phase and the target playback time axis parameters remain stable for several consecutive sampling periods, the backfill trigger flag generation submodule outputs a high-quality backfill trigger flag. The time axis stability includes one or more of the following conditions: the playback direction has not changed, the change in the time scaling ratio is less than a preset threshold, the playback position advance rate fluctuates little, and the current playback position is not in the unstable stage after a zero-crossing switch. When it is detected that the system has exited the stable dwell phase, a new zero-crossing event has occurred, a significant re-acceleration trend has appeared, or the time scaling ratio has changed significantly again in a short period of time, the backfill trigger flag generation submodule cancels the backfill trigger flag. The backfill trigger flag generation submodule outputs the backfill trigger flag and the corresponding trigger or cancellation timing information for subsequent budget allocation and path decision-making.

[0013] Optionally, the temporal reconstruction budget allocation module includes a preview budget generation submodule, a backfill budget generation submodule, and a backfill window determination submodule; The preview budget generation submodule generates a preview budget B_pre that meets real-time output requirements for the preview path based on the current target playback timeline parameters and the current control state S_ctrl. The preview budget B_pre constrains the temporal reconstruction resource investment of the preview path within the current time period, prioritizing the continuity of the display timeline and the low-latency tracking of the push-stick control. The preview budget generation submodule combines playback direction, time scaling ratio, playback position advance rate, output frame period, display cutoff constraints, and the current buffer state to generate a budget set for the preview path. This budget set includes at least one or more of the following: an upper limit on the number of candidate intermediate frames, a generation time limit, a processing complexity level, a processing resolution level, and a total processing allowance. When the buffer fill rate is low or the risk of display cutoff increases, the preview budget generation submodule prioritizes reducing the upper limit on the number of candidate intermediate frames, the processing complexity level, or the processing resolution level. When the buffer state is stable and the display cutoff constraints are met, the preview budget generation submodule allows for an appropriate increase in preview generation capabilities within the budget range. The backfill budget generation submodule generates a backfill budget B_ref for local high-quality temporal reconstruction of the backfill path when the backfill trigger condition is met. When the backfill trigger flag is valid and the system is in a stable dwell phase or meets the stable dwell phase maintenance condition, the backfill budget generation submodule generates a backfill budget B_ref. The backfill budget B_ref includes at least one or more of the following: the number of candidate frames allowed for high-quality reconstruction, the processing complexity level, the processing resolution level, the processing time limit for a single backfill task, and the number of parallel backfill tasks. The establishment and improvement of the backfill budget are based on the premise that the target playback timeline remains stable, and are simultaneously constrained by the system's remaining processing capacity, buffer status, and recent backfill generation efficiency. When the system's processing capacity is insufficient, the risk of display cutoff increases, or recent backfill results are difficult to take effect in a timely manner, the backfill budget generation submodule reduces the backfill budget to avoid interference from high-quality backfill to real-time output. The backfill window determination submodule is used to determine the time window range, i.e., the local backfill window W_ref, based on the generated backfill budget B_ref. The backfill window determination submodule uses the current playback position as a reference, combined with the current playback position advance rate, the current control state S_ctrl, and the backfill budget B_ref, to determine the start and end positions of the local backfill window W_ref. The local backfill window W_ref only covers the interval containing candidate intermediate frames that have not yet reached the display deadline. The local backfill window is determined using one or a combination of the following methods: a symmetrical window extending forward and backward centered on the current playback position; a forward priority window facing the subsequent display interval; a backward compensation window covering candidate frames that have not yet been output and are to be replaced; or an asymmetric adaptive window that dynamically adjusts its boundaries according to the timeline advance direction, display queue structure, and remaining time limit. When the system just enters a stable dwell phase, an initial window length is used; when the stable dwell state continues and the backfill generation efficiency is high, the window is gradually expanded; when the system load increases, the buffer fill rate decreases, or a significant speed fluctuation is detected again, the window length is shortened accordingly.

[0014] Optionally, the dual-path strategy decision module includes a path selection submodule, a display cutoff judgment submodule, and a backup switching submodule; The path selection submodule determines the current frame interpolation path organization method based on the current control state S_ctrl, target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref. When the system is in a fast crossover phase, the path selection submodule prioritizes the path organization method with the preview path as the main component to ensure fast timeline follow-up and low-latency output. When the system is in a near-positioning phase, the path selection submodule maintains the preview path as the main component and establishes conditions for the backfill path to be enabled. When the system is in a stable dwell phase and the backfill trigger flag is valid, the path selection submodule allows the backfill path to be enabled to perform high-quality backfilling of candidate intermediate frames in the local backfill window while maintaining the preview path's responsibility for timely output. The path selection submodule selects the path organization method with the backup path involved based on the current display cutoff risk and system processing capacity. The display cutoff determination submodule performs a display cutoff reachability determination on the current target to be output and the target to be rebuilt within the local backfill window. This determines whether the candidate version can still be generated and used for output before the display cutoff time of the corresponding display timestamp, and whether the replacement version can still be generated before the display cutoff time of the corresponding display timestamp and complete the replacement before the corresponding buffer entry enters the output lock boundary, thus taking effect. The display cutoff determination can be performed at a frame-by-frame or task-by-task granularity, comprehensively considering factors such as the target display timestamp, remaining processing time, current task queue backlog, corresponding path processing load, and output lock boundary. For the candidate intermediate versions to be generated in the preview path... When a frame is determined to be difficult to display on time under the current budget, the display deadline determination submodule triggers a reduction in preview processing intensity or switches to a backup path. For replacement frame versions to be generated in the backfill path, when a determination indicates that they are not expected to be generated before the corresponding display deadline, or are not expected to be replaced before the corresponding buffer entry enters the output locking boundary, the display deadline determination submodule stops investing high-complexity backfill resources into the target and keeps the corresponding candidate version as a valid output version. Within the local backfill window, the display deadline determination submodule allows for reachability determination of different target frames, thereby retaining only backfill tasks that still have practical replacement significance. The backup switching submodule is used to trigger the backup path to take over part or all of the output process when the current path organization cannot meet the display cutoff constraint, or when the system processing capacity is insufficient, the buffer fill rate continues to decline, or the task backlog increases significantly. The backup path adopts one or a combination of the following low-load output methods: reusing existing candidate frames or adjacent output frames, performing low-complexity simplified interpolation, directly outputting the original frame, or adopting other simplified reconstruction methods that can maintain the continuity of the current display timeline. After triggering the backup path, the backup switching submodule keeps the current target playback timeline parameters unchanged and prioritizes output continuity and display timeliness. To avoid the system frequently switching back and forth between the preview path and the backup path near the high load boundary, the backup switching submodule sets a minimum hold period after the backup path is triggered. When the display cutoff risk decreases, the buffer state is restored, and the preview path regains stable output capability, the backup path is then exited and the original path organization is restored.

[0015] Optionally, the two-stage frame interpolation execution module includes a candidate intermediate frame generation submodule, a local backfill triggering submodule, a replacement frame generation submodule, and a display timestamp binding submodule; The candidate intermediate frame generation submodule performs low-latency frame interpolation on the current frame pair to be output based on the preview budget B_pre, generating candidate intermediate frames that meet the real-time display requirements. It combines the target playback timeline parameters and the current display timeline to perform temporal reconstruction on the original frame pair or reference frame pair corresponding to the time to be displayed, and outputs the candidate intermediate frames. The submodule determines the generation order of the candidate intermediate frames based on the order of the target display timestamps or the urgency of the display, prioritizing the reduction of display deadline default risks. The generated candidate intermediate frames are used to meet the continuous output requirements of the current timeline and serve as the base version for subsequent high-quality backfilling. The local backfill trigger submodule is used to filter the set of target frames for high-quality backfilling from the local backfill window W_ref when the conditions for high-quality backfilling are met. When the backfill trigger flag is valid, the system is in a stable dwell phase, and the local backfill window W_ref is determined, the local backfill trigger submodule selects candidate intermediate frames that have generated candidate versions, have not yet been actually output, and have processing margin before the corresponding display deadline as backfillable targets. The local backfill trigger submodule comprehensively considers the target display timestamp, remaining processing time, current task queue backlog, and backfill budget B_ref to sort the backfillable targets and prioritizes the processing of target frames with actual replacement significance. When a new zero-crossing event is detected, the system exits the stable dwell phase, the display deadline risk increases, or a higher-level load reduction condition occurs, the local backfill trigger submodule stops expanding the backfill target set and cancels low-priority backfilling tasks that have not yet started execution. The replacement frame generation submodule performs high-quality frame interpolation on the selected backfillable targets to generate replacement frame versions corresponding to the corresponding candidate intermediate frames. Under the constraint of the backfill budget B_ref, the replacement frame generation submodule performs high-quality temporal reconstruction on the reference frame pairs corresponding to the backfillable targets to obtain replacement frames with the same target temporal position as the original candidate intermediate frames. When generating replacement frames, the replacement frame generation submodule adopts a higher processing complexity level, processing resolution level, or temporal consistency constraint strength than the preview path. The generation of the replacement frames follows the principle of reconstruction at the same time position, that is, the replacement frame and its corresponding candidate intermediate frame are reconstructed for the same target display time. When it is determined that the current backfillable target cannot be generated before the corresponding display deadline, or even if it is generated, it cannot be replaced before the corresponding buffer entry enters the output locking boundary and therefore no longer has practical replacement significance, the replacement frame generation submodule stops the replacement frame generation task corresponding to the current target to avoid ineffective occupation of high-quality processing resources. The display timestamp binding submodule is used to bind candidate intermediate frames and replacement frame versions to their corresponding target display timestamps, and output related information for subsequent versioned buffer management. The display timestamp binding submodule writes at least one or more of the following information for each candidate intermediate frame and replacement frame: display timestamp, target time domain position, version identifier, source path identifier, validity status, and corresponding display deadline. For replacement frame versions, the display timestamp binding submodule maintains that it has the same display timestamp and the same target time domain position as the replaced candidate intermediate frame, so that the subsequent versioned buffer module can perform in-place replacement without changing the playback timeline.

[0016] Optionally, the versioned buffer and seamless replacement output module includes a versioned buffer establishment submodule, a version replacement determination submodule, and a valid version output submodule; The versioned buffer establishment submodule is used to uniformly manage the buffers of candidate intermediate frames and replacement frame versions, and to establish a multi-version organizational structure oriented towards the same display time. The versioned buffer establishment submodule creates buffer entries based on display timestamps, and manages different frame versions corresponding to that display timestamp within each buffer entry. Only candidate versions and replacement versions with the same display timestamp and the same target time domain position are allowed to enter the same buffer entry, ensuring that subsequent replacements occur at the same target display time without changing the current playback timeline. Each buffer entry includes at least one or more of the following information: candidate version slot, replacement version slot, currently valid output version identifier, version priority, version validity status, and output lock status. For historical entries that have been output or have lost their display significance, the versioned buffer establishment submodule releases them to reclaim buffer resources. The version replacement determination submodule is used to determine whether the current replacement version is qualified to replace the corresponding candidate version, and updates the valid output version in the buffer entry when the conditions are met. When a replacement version is qualified to replace, it must meet the following conditions: it has the same display timestamp and the same target time domain position as the original candidate version; it was generated before the corresponding display deadline; and its corresponding buffer entry has not yet entered the output locking boundary. The output locking boundary is set to the display deadline, or a reserved safety boundary earlier than the display deadline, to avoid instability caused by version switching near the actual output time. When the same display timestamp corresponds to multiple replaceable versions, the version replacement determination submodule determines the final valid output version based on version priority, generation completion time, or current valid state. When the replacement version is not generated before the deadline, or the corresponding entry has entered the output locking boundary, the candidate version is kept as a valid output version. The valid version output submodule is used to complete the actual display output according to the current valid output version of each buffer entry in the versioned buffer, in the order of display timestamps. Before the current display timestamp reaches the output time, the valid version output submodule reads the current valid output version from the corresponding buffer entry and sends it to the display output link. When the buffer entry has completed the valid version update before entering the output locking boundary, the updated replacement version is output; otherwise, the original candidate version is output. After the output is completed, for buffer entries that are no longer involved in subsequent display, the valid version output submodule notifies the versioned buffer to release the corresponding resources.

[0017] Optionally, the constraint closed-loop and graded load reduction module includes a constraint index monitoring submodule, a graded load reduction submodule, and a load recovery submodule; The constraint indicator monitoring submodule continuously monitors key constraint indicators during system operation and provides status basis for subsequent graded load reduction and recovery control. These key constraint indicators include at least one or more of the following: display deadline default risk or default rate, processing load of preview and backfill paths, buffer fill rate or fill fluctuation, task queue backlog, backfill result generation efficiency, stability of the current control state S_ctrl, and version switching conflict risk near the output lock boundary. The constraint indicator monitoring submodule's judgment of these constraint indicators is not limited to instantaneous measurements but combines time window statistical results, trend judgments, or segmented threshold rules for comprehensive analysis to avoid over-response caused by short-term peak loads, occasional delays, or instantaneous state fluctuations. The constraint indicator monitoring submodule outputs the current system constraint level or operating status evaluation result for subsequent load reduction and recovery control. The tiered load reduction submodule is used to implement tiered load reduction control based on the current control state S_ctrl and the system's operational risk level when constraint indicators exceed preset ranges. The tiered load reduction submodule prioritizes reducing backfill path resources, then shrinking preview path capabilities, and finally having the safety path take over partial or all output. At lower load reduction levels, the tiered load reduction submodule limits the growth of the backfill budget B_ref, shortens the local backfill window W_ref, stops adding new backfill tasks, or cancels low-priority backfill tasks that have not yet started. At higher load reduction levels, the tiered load reduction submodule further reduces the preview budget B_p. The corresponding candidate frame number limit, processing complexity level, or processing resolution level; when the display cutoff risk continues to increase, the buffer fill rate continues to decrease, or the task backlog cannot be alleviated, the graded load reduction submodule triggers the backup path to take over partial or all output, and locks the system to the minimum runnable configuration when necessary, so as to prioritize the continuous output and display timeliness of the current playback timeline; the load reduction level is executed differently according to whether it is currently in the fast crossing phase, the approach positioning phase, or the stable dwell phase, wherein in the fast crossing phase, the low latency following capability is prioritized to be protected, and in the stable dwell phase, the local high-quality backfill capability is prioritized to be compressed; The load recovery submodule is used to progressively restore the executed load reduction strategy when the system constraint indicators recover to the preset safety range and the recovery conditions are continuously met. The load recovery submodule gradually restores the system capacity in a sequence coordinated with the load reduction, first restoring the stable output margin of the preview path, then restoring the limited backfill budget B_ref and a small local backfill window W_ref, and finally restoring the complete preview and backfill parallel enhancement state. The recovery process can be judged by combining the stability of the buffer state, the trend of the display deadline default risk, the continuous stability of the current control state S_ctrl, and the generation efficiency of the recent backfill results. Among them, the backfill generation efficiency is one of the important thresholds for reopening high-quality backfill capabilities. During the recovery process, if the display deadline risk increases again, the buffer fill rate decreases, the task backlog increases, or the control state switches rapidly again, further recovery is stopped, or the system is rolled back to the previous stable recovery level to avoid the system repeatedly swinging between load reduction and recovery.

[0018] The technical solution provided by this invention includes a follower state acquisition and zero-position calibration module, an operation state determination and state switching module, a playback timeline target generation module, a temporal reconstruction budget allocation module, a dual-path strategy decision module, a dual-stage frame interpolation execution module, a versioned buffer and seamless replacement output module, and a constraint closed-loop and graded load reduction module. The follower state acquisition and zero-position calibration module continuously acquires the follower's position sequence, performs dead-zone calibration on minute jitters near the follower's zero position, and outputs standardized position values, zero-crossing events, and dwell time information. The operation state determination and state switching module extracts displacement change, velocity, acceleration, or velocity based on the position sequence, zero-crossing events, and dwell time information. The system tracks the changing trend and dwell time, and divides the current push rod operation into a rapid crossing phase, an approach positioning phase, and a stable dwell phase according to preset state rules, outputting the current control state; the playback timeline target generation module generates target playback timeline parameters based on the standardized position values ​​and the current control state, and outputs a backfill trigger flag when the current control state is in a stable dwell phase and the target playback timeline parameters remain stable within a certain number of consecutive sampling periods. The target playback timeline parameters include playback direction, time scaling ratio, and the advance rate of the current playback position on the original video timeline; the temporal reconstruction budget allocation module allocates resources based on the current control state, the target playback timeline parameters, display time constraints, and... The system generates a preview budget, a backfill budget, and a partial backfill window during operation. The preview budget constrains resource investment in generating candidate intermediate frames in the preview path, the backfill budget constrains resource investment in generating alternative frame versions in the backfill path, and the partial backfill window limits the allowed time or frame range for high-quality backfilling. The dual-path strategy decision module determines the appropriate frame interpolation path organization among the preview path, backfill path, and safety path based on the current control state, the target playback timeline parameters, the preview budget, the backfill budget, and the partial backfill window, and performs a display cutoff reachability check on the candidate intermediate frames or alternative frame versions. The preview path uses... The system prioritizes generating candidate intermediate frames that meet the current display deadline requirements to ensure continuous timeline and low-latency response. The backfill path generates high-quality replacement frames for candidate intermediate frames within a local backfill window that have not yet reached the display deadline during the stable dwell phase, improving local slow-motion image quality. The safety net path maintains continuous output of the current display timeline in a low-load manner when system processing power is insufficient, display deadline risk increases, or high-quality results cannot take effect on time. The dual-stage frame interpolation execution module performs low-latency frame interpolation on the frames to be output based on the preview budget, generating candidate intermediate frames that meet the current display deadline requirements and binding corresponding display timestamps to the candidate intermediate frames.And when the backfill trigger flag is valid and the current control state is a stable dwell phase, high-quality frame interpolation is performed on candidate intermediate frames within the local backfill window that have not yet reached the corresponding display deadline, so as to generate a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; the versioned buffer and seamless replacement output module is used to establish buffer entries according to the display timestamp, perform versioned management of the candidate intermediate frames and the replacement frame versions, and when the replacement frame version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the current valid output version is changed from The candidate intermediate frames are updated to the replacement frame versions, and the currently valid output versions are output in the order of display timestamps. The constraint closed-loop and hierarchical load reduction module is used to monitor the display deadline default risk, path processing load, buffer filling status, task queue backlog, backfill efficiency, and the stability of the current control state. Based on the monitoring results, it performs hierarchical load reduction and load recovery to prioritize the continuous output and display timeliness of the current playback timeline. While ensuring low-latency timeline tracking, this system improves the display quality of local slow-motion segments during the stable dwell phase and enhances the controllability and stability of the system under actual operating constraints. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of a slow-motion playback real-time frame interpolation control system based on the push rod operation state provided in an embodiment of the present invention. Figure 2 A schematic diagram of the temporal reconstruction budget allocation module provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a dual-path strategy decision module provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of a two-stage frame interpolation execution module provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the versioned buffer and seamless replacement output module provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms unless the context clearly indicates otherwise.

[0023] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0024] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0025] This invention provides a schematic diagram of a real-time frame interpolation control system for slow-motion playback based on the state of a push rod operation, as shown below. Figures 1 to 5 As shown, the system includes: The module includes: push rod status acquisition and zero-position calibration module, operation status determination and status switching module, playback time axis target generation module, time domain reconstruction budget allocation module, dual-path strategy decision module, dual-stage frame interpolation execution module, versioned buffer and seamless replacement output module, and constraint closed loop and graded load reduction module. The push rod status acquisition and zero-position calibration module is used to continuously acquire the position sequence of the push rod, perform dead zone calibration on the small jitters near the zero position of the push rod, and output the standardized position value P(t), zero-crossing event and dwell time information. The operation state determination and state switching module is used to extract displacement change, velocity, acceleration or velocity change trend and dwell time based on the position sequence, zero crossing event and dwell time information, and divide the current push rod operation into rapid crossing phase, approach positioning phase and stable dwell phase according to the preset state rules, and output the current control state S_ctrl. The playback timeline target generation module is used to generate target playback timeline parameters based on the standardized position value P(t) and the current control state S_ctrl, and outputs a backfill trigger flag when the current control state S_ctrl is a stable dwell phase and the target playback timeline parameters remain stable within a number of consecutive sampling periods. The target playback timeline parameters include playback direction, time scaling ratio, and the advance rate of the current playback position on the original video timeline. The temporal reconstruction budget allocation module is used to generate a preview budget B_pre, a backfill budget B_ref, and a local backfill window W_ref based on the current control state S_ctrl, the target playback timeline parameters, the display time limit constraints, and the system running state. The preview budget B_pre is used to constrain the resource investment in generating candidate intermediate frames in the preview path, the backfill budget B_ref is used to constrain the resource investment in generating alternative frame versions in the backfill path, and the local backfill window W_ref is used to limit the time range or frame range that allows high-quality backfill to be performed. The dual-path strategy decision module is used to determine the current frame interpolation path organization method that the system should adopt among the preview path, backfill path, and safety path based on the current control state S_ctrl, the target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref, and to perform a display cutoff reachability judgment on candidate intermediate frames or alternative frame versions. Specifically, the preview path is used to prioritize generating candidate intermediate frames that meet the current display cutoff time requirement to ensure timeline continuity and low-latency response; the backfill path is used to generate high-quality alternative frame versions for candidate intermediate frames within the local backfill window that have not yet reached the display cutoff time during the stable dwell phase to improve local slow-motion image quality; the safety path is used to maintain continuous output of the current display timeline in a low-load manner when the system processing capacity is insufficient, the display cutoff risk is increased, or high-quality results cannot take effect on time. The two-stage frame interpolation execution module is used to perform low-latency frame interpolation on the pair of frames to be output based on the preview budget B_pre, generate candidate intermediate frames that meet the current display deadline requirements, and bind the corresponding display timestamp to the candidate intermediate frames; and when the backfill trigger flag is valid and the current control state S_ctrl is in a stable dwell phase, it performs high-quality frame interpolation on the candidate intermediate frames in the local backfill window W_ref that have not yet reached their corresponding display deadline, so as to generate a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; The versioned buffer and seamless replacement output module is used to establish buffer entries according to the display timestamp, perform versioned management of the candidate intermediate frame and the replacement frame version, and when the replacement frame version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the current valid output version is updated from the candidate intermediate frame to the replacement frame version, and the current valid output version is output according to the display timestamp order. The constraint closed-loop and graded load reduction module is used to monitor the display cutoff default risk, path processing load, buffer filling status, task queue backlog, backfill efficiency, and current control state stability. Based on the monitoring results, it performs graded load reduction and load recovery to prioritize the continuous output and display timeliness of the current playback timeline.

[0026] In this embodiment of the invention, the push rod state acquisition and zero-position calibration module performs dead-zone calibration on the tiny jitters near the zero position of the push rod and outputs standardized position values, zero-crossing events, and dwell time information; the zero-crossing events are used for subsequent stable switching of playback direction, and the dwell time information is used to determine whether the system has entered a stable dwell phase, thereby providing a trigger basis for two-stage frame interpolation and local high-quality backfill; To avoid state jitter near critical conditions, the operation state determination and state switching module sets entry and exit thresholds between adjacent states to form a hysteresis interval. The operation state determination and state switching module outputs the current control state S_ctrl and its switching event, providing a basis for subsequent playback timeline target generation, budget allocation and path decision-making. The playback timeline target generation module sets a freeze zone near the zero position to keep the system stable when switching between normal playback and slow motion playback. When the operation state enters the stable dwell phase, the preset dwell conditions are met, and the target playback timeline parameters remain stable within several consecutive sampling periods, the playback timeline target generation module outputs a high-quality backfill trigger flag to notify subsequent modules to perform higher-quality backfill processing on the frames to be displayed that have not yet reached the display deadline. In the fast-crossing phase, the temporal reconstruction budget allocation module prioritizes allocating the preview budget to meet the low-latency output requirements; in the approach positioning phase, the temporal reconstruction budget allocation module gradually establishes the backfill budget and backfill window; in the stable dwell phase, the temporal reconstruction budget allocation module increases the backfill budget and determines the local backfill window while ensuring that the preview path outputs on time. The dual-path strategy decision module performs an explicit deadline reachability judgment on the current path organization method. This is used to determine whether the candidate version can still be generated and used for output before the explicit deadline of the corresponding explicit timestamp, and whether the alternative version can still be generated before the explicit deadline of the corresponding explicit timestamp and complete the replacement before the corresponding buffer entry enters the output locking boundary and thus take effect. When the judgment result shows that the current path cannot meet the explicit constraints, the backup path is triggered to intervene. When the dual-stage frame interpolation execution module enters a stable dwell phase and continues for more than a preset duration, it initiates background partial backfilling for a preset time window in the vicinity of the current playback position. The partial backfilling re-executes high-quality frame interpolation for candidate intermediate frames within the time window that have not yet reached the display deadline, so as to obtain a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame. Through the dual-stage execution method of previewing first and then backfilling, the system prioritizes ensuring follow-up when the slider moves quickly and prioritizes improving local image quality after the slider stabilizes. The versioned buffer and seamless replacement output module maintains at least two frame versions for the same display time: a candidate version generated by the preview path and a replacement version generated by the backfill path. Each version shares the same display timestamp but has different version identifiers, validity statuses, and priority attributes. The versioned buffer and seamless replacement output module establishes a versioned buffer structure indexed by the display timestamp. When the replacement version corresponding to the current display timestamp is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the valid output version corresponding to the current display timestamp is updated from the candidate version to the replacement version, while maintaining its display timestamp. When the replacement version fails to be generated before the deadline, or the corresponding buffer entry has entered the output locking boundary, the candidate version is maintained as the valid output version for the current display timestamp. During output control, the system selects the highest priority version currently available for display based on the display timestamp and version validity flag. When the system is in a fast pass-through phase, the constraint-loop and graded load reduction module prioritizes reducing the backfill budget, compressing the backfill window, or temporarily disabling the backfill path to ensure timeline tracking and low-latency output. When the system is in a stable dwell phase, the constraint-loop and graded load reduction module prioritizes shortening the local backfill window, reducing the backfill complexity level, or reducing the number of local frames to be rebuilt. When necessary, the constraint-loop and graded load reduction module switches to the safety path strategy to avoid buffer underflow or continuous accumulation of display defaults. When the system load recovers and continues to meet the preset conditions, the constraint-loop and graded load reduction module gradually restores the backfill budget and local backfill window length in the reverse order of load reduction.

[0027] In this embodiment of the invention, the push rod state acquisition and zero-position calibration module includes a position acquisition and standardization submodule and a zero-position dead zone calibration and event output submodule; The position acquisition and standardization submodule continuously acquires the displacement information of the push rod and converts the acquired raw displacement into a standardized position value that can be used for subsequent control. The submodule obtains the raw position sample values ​​of the push rod and normalizes them according to the push rod's mechanical stroke range to obtain the standardized position value P(t). The submodule outputs a position sequence arranged chronologically and adds sampling timing information to each sample value for subsequent modules to calculate dynamic characteristics such as displacement change, velocity, acceleration, and dwell time. The submodule performs basic stabilization processing on the raw sample values ​​to suppress the impact of instantaneous spike noise or abnormal jumps on subsequent control. The final output of the position acquisition and standardization submodule includes at least: the raw position value, the standardized position value P(t), and the corresponding sampling timing information. The zero-position dead zone calibration and event output submodule is used to perform dead zone calibration on minute jitters near the zero position of the push rod, and outputs control event information related to the zero position based on the standardized position sequence. The zero-position dead zone calibration and event output submodule establishes a zero-position dead zone around the push rod zero position, the boundary of which is determined based on a preset threshold, equipment calibration results, or historical fluctuation statistics. The zero-position dead zone is set using a fixed threshold method or corrected using an adaptive method based on historical fluctuation statistics. When the standardized position value P(t) falls within the zero-position dead zone range, the zero-position dead zone calibration and event output submodule outputs a zero-position locked state. When the standardized position value P(t) moves from one side of the zero-position dead zone... When crossing to the other side, the zero-position dead zone calibration and event output submodule outputs a zero-crossing event signal; when the standardized position value P(t) remains in a low-change state near the zero position or the current position and continues for more than a preset duration, the zero-position dead zone calibration and event output submodule accumulates dwell time information; to avoid frequent entry and exit jitter near the zero-position boundary, the zero-position dead zone calibration and event output submodule sets an entry threshold and an exit threshold at the zero-position dead zone boundary, so that the establishment and release of the zero-position locking state have hysteresis characteristics; the final output of the zero-position dead zone calibration and event output submodule includes at least: zero-position locking state, zero-crossing event, dwell time value, and timing information corresponding to the event.

[0028] In this embodiment of the invention, the operation state determination and state switching module includes a dynamic feature extraction submodule, a state determination submodule, and a hysteresis switching and state output submodule. The dynamic feature extraction submodule receives standardized position values ​​P(t), zero-position locking status, zero-crossing events, and dwell time information, and extracts dynamic features for operation state determination based on continuous position sequences. The submodule maintains a position sequence or sliding time window updated chronologically, and calculates at least one or more of the following dynamic features: displacement change, velocity, acceleration or velocity change trend, dwell time, position fluctuation amplitude, and zero-position related state information. To improve the stability of dynamic features, the submodule smooths the velocity and acceleration to suppress interference from mechanical micro-vibrations, sampling noise, or short-term abnormal jumps. The submodule outputs a set of feature parameters for state determination, which are then used for subsequent state determination and state switching.

[0029] The state determination submodule is used to determine the current push rod operation state based on the feature parameters output by the dynamic feature extraction submodule, and classify it into one of the following: rapid crossing phase, approach positioning phase, and stable dwell phase. When the absolute value of the push rod's movement speed is higher than a preset first speed threshold and the position change is continuously large within the current time window, it is determined to be a rapid crossing phase. When the absolute value of the push rod's movement speed is less than the first speed threshold but greater than a preset second speed threshold, and the speed change trend shows a decreasing trend, it is determined to be an approach positioning phase. When the position change of the push rod within the current time window is less than a preset position change threshold and the duration is greater than a preset dwell time threshold, it is determined to be a stable dwell phase. The first speed threshold, the second speed threshold, the preset position change threshold, and the preset dwell time threshold are set according to the equipment calibration results, the push rod's mechanical stroke, the position sampling resolution, or the historical operation statistics results.

[0030] The hysteresis switching and state output submodule is used to stabilize the state determination result and output the final effective control state. The hysteresis switching and state output submodule sets entry and exit thresholds between adjacent states to give the state switching hysteresis characteristics, avoiding frequent state jumps due to short-term fluctuations near the threshold boundaries. The hysteresis switching and state output submodule combines zero-position locking state and zero-crossing events to constrain state switching. When the system has entered a stable dwell phase, if a new zero-crossing event, continuous large position changes, or a significant re-acceleration trend is detected, it exits the stable dwell phase and switches to the approach positioning phase or rapid crossing phase. The hysteresis switching and state output submodule ultimately outputs the current control state S_ctrl and the corresponding state switching event information for subsequent playback timeline target generation, budget allocation, path selection, and load reduction recovery control.

[0031] In this embodiment of the invention, the playback timeline target generation module includes a timeline parameter generation submodule and a backfill trigger flag generation submodule; The timeline parameter generation submodule receives the standardized position value P(t), zero-position locking status, zero-crossing event, and current control state S_ctrl, and generates target playback timeline parameters based on the above inputs. The target playback timeline parameters include at least the playback direction, time scaling ratio, and the current playback position's advance rate on the original video timeline. The timeline parameter generation submodule generates the playback direction and time scaling ratio based on the magnitude and sign of the standardized position value P(t), and sets a freeze zone near the zero position. When the standardized position value P(t) falls within the freeze zone, the timeline parameter generation submodule maintains the current valid timeline parameters or smoothly transitions to normal playback according to preset rules to avoid frequent changes in playback direction or advance rate due to slight oscillations of the push bar near the zero position. Upper and lower limits are set for the time scaling ratio and playback position advance rate to limit abrupt changes or excessively slow advance of the timeline caused by abnormal inputs. The timeline parameter generation submodule outputs the current playback direction, time scaling ratio, playback position advance rate, and corresponding timing information. The backfill trigger flag generation submodule combines the current control state S_ctrl with the target playback time axis parameters to determine whether the system meets the basic conditions for entering the high-quality local backfill stage, and outputs a backfill trigger flag when the conditions are met. When the system is in a stable dwell phase and the target playback time axis parameters remain stable for several consecutive sampling periods, the backfill trigger flag generation submodule outputs a high-quality backfill trigger flag. The time axis stability includes one or more of the following conditions: the playback direction has not changed, the change in the time scaling ratio is less than a preset threshold, the playback position advance rate fluctuates little, and the current playback position is not in the unstable stage after a zero-crossing switch. When it is detected that the system has exited the stable dwell phase, a new zero-crossing event has occurred, a significant re-acceleration trend has appeared, or the time scaling ratio has changed significantly again in a short period of time, the backfill trigger flag generation submodule cancels the backfill trigger flag. The backfill trigger flag generation submodule outputs the backfill trigger flag and the corresponding trigger or cancellation timing information for subsequent budget allocation and path decision-making.

[0032] In this embodiment of the invention, the temporal reconstruction budget allocation module includes a preview budget generation submodule, a backfill budget generation submodule, and a backfill window determination submodule; The preview budget generation submodule generates a preview budget B_pre that meets real-time output requirements for the preview path based on the current target playback timeline parameters and the current control state S_ctrl. The preview budget B_pre constrains the temporal reconstruction resource investment of the preview path within the current time period, prioritizing the continuity of the display timeline and the low-latency tracking of the push-stick control. The preview budget generation submodule combines playback direction, time scaling ratio, playback position advance rate, output frame period, display cutoff constraints, and the current buffer state to generate a budget set for the preview path. This budget set includes at least one or more of the following: an upper limit on the number of candidate intermediate frames, a generation time limit, a processing complexity level, a processing resolution level, and a total processing allowance. When the buffer fill rate is low or the risk of display cutoff increases, the preview budget generation submodule prioritizes reducing the upper limit on the number of candidate intermediate frames, the processing complexity level, or the processing resolution level. When the buffer state is stable and the display cutoff constraints are met, the preview budget generation submodule allows for an appropriate increase in preview generation capabilities within the budget range. The backfill budget generation submodule generates a backfill budget B_ref for local high-quality temporal reconstruction of the backfill path when the backfill trigger condition is met. When the backfill trigger flag is valid and the system is in a stable dwell phase or meets the stable dwell phase maintenance condition, the backfill budget generation submodule generates a backfill budget B_ref. The backfill budget B_ref includes at least one or more of the following: the number of candidate frames allowed for high-quality reconstruction, the processing complexity level, the processing resolution level, the processing time limit for a single backfill task, and the number of parallel backfill tasks. The establishment and improvement of the backfill budget are based on the premise that the target playback timeline remains stable, and are simultaneously constrained by the system's remaining processing capacity, buffer status, and recent backfill generation efficiency. When the system's processing capacity is insufficient, the risk of display cutoff increases, or recent backfill results are difficult to take effect in a timely manner, the backfill budget generation submodule reduces the backfill budget to avoid interference from high-quality backfill to real-time output. The backfill window determination submodule is used to determine the time window range, i.e., the local backfill window W_ref, based on the generated backfill budget B_ref. The backfill window determination submodule uses the current playback position as a reference, combined with the current playback position advance rate, the current control state S_ctrl, and the backfill budget B_ref, to determine the start and end positions of the local backfill window W_ref. The local backfill window W_ref only covers the interval containing candidate intermediate frames that have not yet reached the display deadline. The local backfill window is determined using one or a combination of the following methods: a symmetrical window extending forward and backward centered on the current playback position; a forward priority window facing the subsequent display interval; a backward compensation window covering candidate frames that have not yet been output and are to be replaced; or an asymmetric adaptive window that dynamically adjusts its boundaries according to the timeline advance direction, display queue structure, and remaining time limit. When the system just enters a stable dwell phase, an initial window length is used; when the stable dwell state continues and the backfill generation efficiency is high, the window is gradually expanded; when the system load increases, the buffer fill rate decreases, or a significant speed fluctuation is detected again, the window length is shortened accordingly.

[0033] In this embodiment of the invention, the dual-path strategy decision module includes a path selection submodule, a display cutoff judgment submodule, and a backup switching submodule; The path selection submodule determines the current frame interpolation path organization method based on the current control state S_ctrl, target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref. When the system is in a fast crossover phase, the path selection submodule prioritizes the path organization method with the preview path as the main component to ensure fast timeline follow-up and low-latency output. When the system is in a near-positioning phase, the path selection submodule maintains the preview path as the main component and establishes conditions for the backfill path to be enabled. When the system is in a stable dwell phase and the backfill trigger flag is valid, the path selection submodule allows the backfill path to be enabled to perform high-quality backfilling of candidate intermediate frames in the local backfill window while maintaining the preview path's responsibility for timely output. The path selection submodule selects the path organization method with the backup path involved based on the current display cutoff risk and system processing capacity. The display cutoff determination submodule performs a display cutoff reachability determination on the current target to be output and the target to be rebuilt within the local backfill window. This determines whether the candidate version can still be generated and used for output before the display cutoff time of the corresponding display timestamp, and whether the replacement version can still be generated before the display cutoff time of the corresponding display timestamp and complete the replacement before the corresponding buffer entry enters the output lock boundary, thus taking effect. The display cutoff determination can be performed at a frame-by-frame or task-by-task granularity, comprehensively considering factors such as the target display timestamp, remaining processing time, current task queue backlog, corresponding path processing load, and output lock boundary. For the candidate intermediate versions to be generated in the preview path... When a frame is determined to be difficult to display on time under the current budget, the display deadline determination submodule triggers a reduction in preview processing intensity or switches to a backup path. For replacement frame versions to be generated in the backfill path, when a determination indicates that they are not expected to be generated before the corresponding display deadline, or are not expected to be replaced before the corresponding buffer entry enters the output locking boundary, the display deadline determination submodule stops investing high-complexity backfill resources into the target and keeps the corresponding candidate version as a valid output version. Within the local backfill window, the display deadline determination submodule allows for reachability determination of different target frames, thereby retaining only backfill tasks that still have practical replacement significance. The backup switching submodule is used to trigger the backup path to take over part or all of the output process when the current path organization cannot meet the display cutoff constraint, or when the system processing capacity is insufficient, the buffer fill rate continues to decline, or the task backlog increases significantly. The backup path adopts one or a combination of the following low-load output methods: reusing existing candidate frames or adjacent output frames, performing low-complexity simplified interpolation, directly outputting the original frame, or adopting other simplified reconstruction methods that can maintain the continuity of the current display timeline. After triggering the backup path, the backup switching submodule keeps the current target playback timeline parameters unchanged and prioritizes output continuity and display timeliness. To avoid the system frequently switching back and forth between the preview path and the backup path near the high load boundary, the backup switching submodule sets a minimum hold period after the backup path is triggered. When the display cutoff risk decreases, the buffer state is restored, and the preview path regains stable output capability, the backup path is then exited and the original path organization is restored.

[0034] In this embodiment of the invention, the two-stage frame interpolation execution module includes a candidate intermediate frame generation submodule, a local backfill triggering submodule, a replacement frame generation submodule, and a display timestamp binding submodule; The candidate intermediate frame generation submodule performs low-latency frame interpolation on the current frame pair to be output based on the preview budget B_pre, generating candidate intermediate frames that meet the real-time display requirements. It combines the target playback timeline parameters and the current display timeline to perform temporal reconstruction on the original frame pair or reference frame pair corresponding to the time to be displayed, and outputs the candidate intermediate frames. The submodule determines the generation order of the candidate intermediate frames based on the order of the target display timestamps or the urgency of the display, prioritizing the reduction of display deadline default risks. The generated candidate intermediate frames are used to meet the continuous output requirements of the current timeline and serve as the base version for subsequent high-quality backfilling. The local backfill trigger submodule is used to filter the set of target frames for high-quality backfilling from the local backfill window W_ref when the conditions for high-quality backfilling are met. When the backfill trigger flag is valid, the system is in a stable dwell phase, and the local backfill window W_ref is determined, the local backfill trigger submodule selects candidate intermediate frames that have generated candidate versions, have not yet been actually output, and have processing margin before the corresponding display deadline as backfillable targets. The local backfill trigger submodule comprehensively considers the target display timestamp, remaining processing time, current task queue backlog, and backfill budget B_ref to sort the backfillable targets and prioritizes the processing of target frames with actual replacement significance. When a new zero-crossing event is detected, the system exits the stable dwell phase, the display deadline risk increases, or a higher-level load reduction condition occurs, the local backfill trigger submodule stops expanding the backfill target set and cancels low-priority backfilling tasks that have not yet started execution. The replacement frame generation submodule performs high-quality frame interpolation on the selected backfillable targets to generate replacement frame versions corresponding to the corresponding candidate intermediate frames. Under the constraint of the backfill budget B_ref, the replacement frame generation submodule performs high-quality temporal reconstruction on the reference frame pairs corresponding to the backfillable targets to obtain replacement frames with the same target temporal position as the original candidate intermediate frames. When generating replacement frames, the replacement frame generation submodule adopts a higher processing complexity level, processing resolution level, or temporal consistency constraint strength than the preview path. The generation of the replacement frames follows the principle of reconstruction at the same time position, that is, the replacement frame and its corresponding candidate intermediate frame are reconstructed for the same target display time. When it is determined that the current backfillable target cannot be generated before the corresponding display deadline, or even if it is generated, it cannot be replaced before the corresponding buffer entry enters the output locking boundary and therefore no longer has practical replacement significance, the replacement frame generation submodule stops the replacement frame generation task corresponding to the current target to avoid ineffective occupation of high-quality processing resources. The display timestamp binding submodule is used to bind candidate intermediate frames and replacement frame versions to their corresponding target display timestamps, and output related information for subsequent versioned buffer management. The display timestamp binding submodule writes at least one or more of the following information for each candidate intermediate frame and replacement frame: display timestamp, target time domain position, version identifier, source path identifier, validity status, and corresponding display deadline. For replacement frame versions, the display timestamp binding submodule maintains that it has the same display timestamp and the same target time domain position as the replaced candidate intermediate frame, so that the subsequent versioned buffer module can perform in-place replacement without changing the playback timeline.

[0035] In this embodiment of the invention, the versioned buffer and seamless replacement output module includes a versioned buffer establishment submodule, a version replacement determination submodule, and a valid version output submodule; The versioned buffer establishment submodule is used to uniformly manage the buffers of candidate intermediate frames and replacement frame versions, and to establish a multi-version organizational structure oriented towards the same display time. The versioned buffer establishment submodule creates buffer entries based on display timestamps, and manages different frame versions corresponding to that display timestamp within each buffer entry. Only candidate versions and replacement versions with the same display timestamp and the same target time domain position are allowed to enter the same buffer entry, ensuring that subsequent replacements occur at the same target display time without changing the current playback timeline. Each buffer entry includes at least one or more of the following information: candidate version slot, replacement version slot, currently valid output version identifier, version priority, version validity status, and output lock status. For historical entries that have been output or have lost their display significance, the versioned buffer establishment submodule releases them to reclaim buffer resources. The version replacement determination submodule is used to determine whether the current replacement version is qualified to replace the corresponding candidate version, and updates the valid output version in the buffer entry when the conditions are met. When a replacement version is qualified to replace, it must meet the following conditions: it has the same display timestamp and the same target time domain position as the original candidate version; it was generated before the corresponding display deadline; and its corresponding buffer entry has not yet entered the output locking boundary. The output locking boundary is set to the display deadline, or a reserved safety boundary earlier than the display deadline, to avoid instability caused by version switching near the actual output time. When the same display timestamp corresponds to multiple replaceable versions, the version replacement determination submodule determines the final valid output version based on version priority, generation completion time, or current valid state. When the replacement version is not generated before the deadline, or the corresponding entry has entered the output locking boundary, the candidate version is kept as a valid output version. The valid version output submodule is used to complete the actual display output according to the current valid output version of each buffer entry in the versioned buffer, in the order of display timestamps. Before the current display timestamp reaches the output time, the valid version output submodule reads the current valid output version from the corresponding buffer entry and sends it to the display output link. When the buffer entry has completed the valid version update before entering the output locking boundary, the updated replacement version is output; otherwise, the original candidate version is output. After the output is completed, for buffer entries that are no longer involved in subsequent display, the valid version output submodule notifies the versioned buffer to release the corresponding resources.

[0036] In this embodiment of the invention, the constraint closed-loop and graded load reduction module includes a constraint index monitoring submodule, a graded load reduction submodule, and a load recovery submodule; The constraint indicator monitoring submodule continuously monitors key constraint indicators during system operation and provides status basis for subsequent graded load reduction and recovery control. These key constraint indicators include at least one or more of the following: display deadline default risk or default rate, processing load of preview and backfill paths, buffer fill rate or fill fluctuation, task queue backlog, backfill result generation efficiency, stability of the current control state S_ctrl, and version switching conflict risk near the output lock boundary. The constraint indicator monitoring submodule's judgment of these constraint indicators is not limited to instantaneous measurements but combines time window statistical results, trend judgments, or segmented threshold rules for comprehensive analysis to avoid over-response caused by short-term peak loads, occasional delays, or instantaneous state fluctuations. The constraint indicator monitoring submodule outputs the current system constraint level or operating status evaluation result for subsequent load reduction and recovery control. The tiered load reduction submodule is used to implement tiered load reduction control based on the current control state S_ctrl and the system's operational risk level when constraint indicators exceed preset ranges. The tiered load reduction submodule prioritizes reducing backfill path resources, then shrinking preview path capabilities, and finally having the safety path take over partial or all output. At lower load reduction levels, the tiered load reduction submodule limits the growth of the backfill budget B_ref, shortens the local backfill window W_ref, stops adding new backfill tasks, or cancels low-priority backfill tasks that have not yet started. At higher load reduction levels, the tiered load reduction submodule further reduces the preview budget B_p. The corresponding candidate frame number limit, processing complexity level, or processing resolution level; when the display cutoff risk continues to increase, the buffer fill rate continues to decrease, or the task backlog cannot be alleviated, the graded load reduction submodule triggers the backup path to take over partial or all output, and locks the system to the minimum runnable configuration when necessary, so as to prioritize the continuous output and display timeliness of the current playback timeline; the load reduction level is executed differently according to whether it is currently in the fast crossing phase, the approach positioning phase, or the stable dwell phase, wherein in the fast crossing phase, the low latency following capability is prioritized to be protected, and in the stable dwell phase, the local high-quality backfill capability is prioritized to be compressed; The load recovery submodule is used to progressively restore the executed load reduction strategy when the system constraint indicators recover to the preset safety range and the recovery conditions are continuously met. The load recovery submodule gradually restores the system capacity in a sequence coordinated with the load reduction, first restoring the stable output margin of the preview path, then restoring the limited backfill budget B_ref and a small local backfill window W_ref, and finally restoring the complete preview and backfill parallel enhancement state. The recovery process can be judged by combining the stability of the buffer state, the trend of the display deadline default risk, the continuous stability of the current control state S_ctrl, and the generation efficiency of the recent backfill results. Among them, the backfill generation efficiency is one of the important thresholds for reopening high-quality backfill capabilities. During the recovery process, if the display deadline risk increases again, the buffer fill rate decreases, the task backlog increases, or the control state switches rapidly again, further recovery is stopped, or the system is rolled back to the previous stable recovery level to avoid the system repeatedly swinging between load reduction and recovery.

[0037] The core workflow of this invention is as follows: After the operator pushes the push rod, the push rod status acquisition and zero-position calibration module continuously outputs standardized position values, zero-crossing events, and dwell time information; the operation status determination and state switching module determines whether the current state is in the rapid crossing phase, the approach positioning phase, or the stable dwell phase based on the push rod's dynamic characteristics; the playback timeline target generation module generates target playback timeline parameters based on the push rod's absolute position and outputs a high-quality backfill trigger flag when conditions are met; the time domain reconstruction budget allocation module generates preview budget and backfill budget based on the current control state, time scaling ratio, buffer state, and processing time limit; the dual-path strategy decision module selects the preview path strategy and the backfill strategy accordingly. The system employs either a fill-in path strategy or a backup path strategy. A two-stage frame interpolation module first generates candidate intermediate frames that meet real-time display requirements, then performs high-quality backfilling on candidate intermediate frames within a local time window that have not yet reached the display deadline. A versioned buffering and seamless replacement output module manages different frame versions under the same display timestamp and target time domain location, and seamlessly switches when the replacement version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary. A constraint closed-loop and graded load reduction module continuously monitors the deadline default rate, buffer fluctuations, and backfilling efficiency, and performs graded load reduction and recovery control based on the current control status.

[0038] The core of this invention lies in zero-position stabilization processing and operation state determination of the push rod input, generating preview budget and backfill budget around the target playback timeline, and performing high-quality backfilling on candidate intermediate frames within a local window that have not yet reached the display deadline, based on the real-time preview output. Furthermore, through versioned buffering based on the same display timestamp and the same target time domain position, seamless replacement output of the alternative version is achieved when the alternative version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary. This ensures low-latency timeline tracking while improving the display quality of local slow-motion segments during the stable dwell phase and enhancing the controllability and stability of the system under actual operational constraints.

[0039] Compared with the prior art, the present invention has the following technical effects: 1. This invention can perform phased control of the slow-motion playback process according to the state of the push rod operation. When the push rod moves quickly, it prioritizes ensuring low-latency time axis following. After the push rod stabilizes, it performs high-quality reconstruction of local segments, thus balancing real-time responsiveness and viewing quality. 2. By limiting high-quality backfilling to a local backfilling window and prioritizing its application to the target time before the display deadline, this invention reduces the computational waste caused by continuously performing high-complexity frame interpolation throughout the entire playback process, thereby improving resource utilization efficiency. 3. This invention establishes a versioned buffer structure based on the same display timestamp, enabling subsequently generated high-quality results to replace the target time that has not yet been displayed without disrupting the continuity of the current playback timeline, thereby improving the display effect of local slow-motion segments. 4. This invention improves the output stability of the system during real-time operation by incorporating factors such as display deadline, buffer status, task backlog, backfilling result generation efficiency, and operational stability into closed-loop control, and by combining a graded load reduction and recovery mechanism, thereby reducing the occurrence of display failures and frequent jitter switching.

[0040] The technical solution provided by this invention includes a follower state acquisition and zero-position calibration module, an operation state determination and state switching module, a playback timeline target generation module, a temporal reconstruction budget allocation module, a dual-path strategy decision module, a dual-stage frame interpolation execution module, a versioned buffer and seamless replacement output module, and a constraint closed-loop and graded load reduction module. The follower state acquisition and zero-position calibration module continuously acquires the follower's position sequence, performs dead-zone calibration on minute jitters near the follower's zero position, and outputs standardized position values, zero-crossing events, and dwell time information. The operation state determination and state switching module extracts displacement change, velocity, acceleration, or velocity based on the position sequence, zero-crossing events, and dwell time information. The system tracks the changing trend and dwell time, and divides the current push rod operation into a rapid crossing phase, an approach positioning phase, and a stable dwell phase according to preset state rules, outputting the current control state; the playback timeline target generation module generates target playback timeline parameters based on the standardized position values ​​and the current control state, and outputs a backfill trigger flag when the current control state is in a stable dwell phase and the target playback timeline parameters remain stable within a certain number of consecutive sampling periods. The target playback timeline parameters include playback direction, time scaling ratio, and the advance rate of the current playback position on the original video timeline; the temporal reconstruction budget allocation module allocates resources based on the current control state, the target playback timeline parameters, display time constraints, and... The system generates a preview budget, a backfill budget, and a partial backfill window during operation. The preview budget constrains resource investment in generating candidate intermediate frames in the preview path, the backfill budget constrains resource investment in generating alternative frame versions in the backfill path, and the partial backfill window limits the allowed time or frame range for high-quality backfilling. The dual-path strategy decision module determines the appropriate frame interpolation path organization among the preview path, backfill path, and safety path based on the current control state, the target playback timeline parameters, the preview budget, the backfill budget, and the partial backfill window, and performs a display cutoff reachability check on the candidate intermediate frames or alternative frame versions. The preview path uses... The system prioritizes generating candidate intermediate frames that meet the current display deadline requirements to ensure continuous timeline and low-latency response. The backfill path generates high-quality replacement frames for candidate intermediate frames within a local backfill window that have not yet reached the display deadline during the stable dwell phase, improving local slow-motion image quality. The safety net path maintains continuous output of the current display timeline in a low-load manner when system processing power is insufficient, display deadline risk increases, or high-quality results cannot take effect on time. The dual-stage frame interpolation execution module performs low-latency frame interpolation on the frames to be output based on the preview budget, generating candidate intermediate frames that meet the current display deadline requirements and binding corresponding display timestamps to the candidate intermediate frames.And when the backfill trigger flag is valid and the current control state is a stable dwell phase, high-quality frame interpolation is performed on candidate intermediate frames within the local backfill window that have not yet reached the corresponding display deadline, so as to generate a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; the versioned buffer and seamless replacement output module is used to establish buffer entries according to the display timestamp, perform versioned management of the candidate intermediate frames and the replacement frame versions, and when the replacement frame version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the current valid output version is changed from The candidate intermediate frames are updated to the replacement frame versions, and the currently valid output versions are output in the order of display timestamps. The constraint closed-loop and hierarchical load reduction module is used to monitor the display deadline default risk, path processing load, buffer filling status, task queue backlog, backfill efficiency, and the stability of the current control state. Based on the monitoring results, it performs hierarchical load reduction and load recovery to prioritize the continuous output and display timeliness of the current playback timeline. While ensuring low-latency timeline tracking, this system improves the display quality of local slow-motion segments during the stable dwell phase and enhances the controllability and stability of the system under actual operating constraints.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time frame interpolation control system for slow-motion playback based on the state of a push rod operation, characterized in that, The system includes: a push rod status acquisition and zero-position calibration module, an operation status determination and status switching module, a playback timeline target generation module, a time domain reconstruction budget allocation module, a dual-path strategy decision module, a dual-stage frame interpolation execution module, a versioned buffer and seamless replacement output module, and a constraint closed loop and graded load reduction module. The push rod status acquisition and zero-position calibration module is used to continuously acquire the position sequence of the push rod, perform dead zone calibration on the small jitters near the zero position of the push rod, and output the standardized position value P(t), zero-crossing event and dwell time information. The operation state determination and state switching module is used to extract displacement change, velocity, acceleration or velocity change trend and dwell time based on the position sequence, zero crossing event and dwell time information, and divide the current push rod operation into rapid crossing phase, approach positioning phase and stable dwell phase according to the preset state rules, and output the current control state S_ctrl. The playback timeline target generation module is used to generate target playback timeline parameters based on the standardized position value P(t) and the current control state S_ctrl, and outputs a backfill trigger flag when the current control state S_ctrl is a stable dwell phase and the target playback timeline parameters remain stable within a number of consecutive sampling periods. The target playback timeline parameters include playback direction, time scaling ratio, and the advance rate of the current playback position on the original video timeline. The temporal reconstruction budget allocation module is used to generate a preview budget B_pre, a backfill budget B_ref, and a local backfill window W_ref based on the current control state S_ctrl, the target playback timeline parameters, the display time limit constraints, and the system running state. The preview budget B_pre is used to constrain the resource investment in generating candidate intermediate frames in the preview path, the backfill budget B_ref is used to constrain the resource investment in generating alternative frame versions in the backfill path, and the local backfill window W_ref is used to limit the time range or frame range that allows high-quality backfill to be performed. The dual-path strategy decision module is used to determine the current frame interpolation path organization method that the system should adopt among the preview path, backfill path, and safety path based on the current control state S_ctrl, the target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref, and to perform a display deadline reachability judgment on candidate intermediate frames or alternative frame versions. Specifically, the preview path is used to prioritize generating candidate intermediate frames that meet the current display deadline requirements to ensure timeline continuity and low-latency response; the backfill path is used to generate high-quality alternative frame versions for candidate intermediate frames within the local backfill window that have not yet reached the display deadline during the stable dwell phase. To improve local slow-motion image quality; the backup path is used to maintain continuous output of the current display timeline in a low-load manner when the system processing capacity is insufficient, the display deadline risk is increased, or high-quality results cannot take effect on time; the dual-path strategy decision module performs a display deadline reachability judgment on the current path organization method, which is used to determine whether the candidate version can still be generated and used for output before the display deadline of the corresponding display timestamp, and to determine whether the alternative version can still be generated before the display deadline of the corresponding display timestamp and complete the replacement and take effect before the corresponding buffer entry enters the output locking boundary; when the judgment result shows that the current path cannot meet the display constraints, the backup path is triggered to intervene; The two-stage frame interpolation execution module is used to perform low-latency frame interpolation on the pair of frames to be output based on the preview budget B_pre, generate candidate intermediate frames that meet the current display deadline requirements, and bind the corresponding display timestamp to the candidate intermediate frames; and when the backfill trigger flag is valid and the current control state S_ctrl is in a stable resident phase, it performs high-quality frame interpolation on the candidate intermediate frames in the local backfill window W_ref that have not yet reached the corresponding display deadline, so as to generate a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; The versioned buffer and seamless replacement output module is used to establish buffer entries according to the display timestamp, perform versioned management of the candidate intermediate frame and the replacement frame version, and when the replacement frame version is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the current valid output version is updated from the candidate intermediate frame to the replacement frame version, and the current valid output version is output according to the display timestamp order. The constraint closed-loop and graded load reduction module is used to monitor the display cutoff default risk, path processing load, buffer filling status, task queue backlog, backfill efficiency, and current control state stability. Based on the monitoring results, it performs graded load reduction and load recovery to prioritize the continuous output and display timeliness of the current playback timeline.

2. The system according to claim 1, characterized in that, The zero-crossing event is used for subsequent stable switching of playback direction, and the dwell time information is used to determine whether the system has entered a stable dwell phase, thereby providing a trigger basis for dual-stage frame interpolation and local high-quality backfilling. To avoid state jitter near critical conditions, the operation state determination and state switching module sets entry and exit thresholds between adjacent states to form a hysteresis interval. The operation state determination and state switching module outputs the current control state S_ctrl and its switching event, providing a basis for subsequent playback timeline target generation, budget allocation and path decision-making. The playback timeline target generation module sets a freeze zone near the zero position to keep the system stable when switching between normal playback and slow motion playback. When the operation state enters the stable dwell phase, the preset dwell conditions are met, and the target playback timeline parameters remain stable within several consecutive sampling periods, the playback timeline target generation module outputs a high-quality backfill trigger flag to notify subsequent modules to perform higher-quality backfill processing on the frames to be displayed that have not yet reached the display deadline. In the fast-crossing phase, the temporal reconstruction budget allocation module prioritizes allocating the preview budget to meet the low-latency output requirements; in the approach positioning phase, the temporal reconstruction budget allocation module gradually establishes the backfill budget and backfill window; in the stable dwell phase, the temporal reconstruction budget allocation module increases the backfill budget and determines the local backfill window while ensuring that the preview path outputs on time. When the operation state enters a stable dwell phase and continues for more than a preset duration, the dual-stage frame interpolation execution module starts background local backfilling for a preset time window in the vicinity of the current playback position. The local backfilling re-executes high-quality frame interpolation for candidate intermediate frames within the time window that have not yet reached the display deadline, so as to obtain a replacement frame version with the same display timestamp and the same target time domain position as the corresponding candidate intermediate frame; through the two-stage execution method of previewing first and then backfilling, the system prioritizes ensuring the following performance when the slider moves quickly, and prioritizes improving the local image quality after the slider stabilizes. The versioned buffer and seamless replacement output module maintains at least two frame versions for the same display time: a candidate version generated by the preview path and an alternative version generated by the backfill path; each version shares the same display timestamp, but has different version identifiers, validity statuses, and priority attributes; the versioned buffer and seamless replacement output module establishes a versioned buffer structure indexed by the display timestamp. If the alternative version corresponding to the current display timestamp is generated before the corresponding display deadline and the corresponding buffer entry has not entered the output locking boundary, the valid output version corresponding to the current display timestamp will be updated from the candidate version to the alternative version, while keeping its display timestamp unchanged. If the alternative version fails to be generated before the deadline, or the corresponding buffer entry has entered the output locking boundary, the candidate version will be kept as the valid output version for the current display timestamp. During output control, the system selects the highest priority version currently available for display based on the display timestamp and version validity flag. When the system is in a fast pass-through phase, the constraint-loop and graded load reduction module prioritizes reducing the backfill budget, compressing the backfill window, or temporarily disabling the backfill path to ensure timeline tracking and low-latency output. When the system is in a stable dwell phase, the constraint-loop and graded load reduction module prioritizes shortening the local backfill window, reducing the backfill complexity level, or reducing the number of local frames to be rebuilt. When necessary, the constraint-loop and graded load reduction module switches to the safety path strategy to avoid buffer underflow or continuous accumulation of display defaults. When the system load recovers and continues to meet the preset conditions, the constraint-loop and graded load reduction module gradually restores the backfill budget and local backfill window length in the reverse order of load reduction.

3. The system according to claim 2, characterized in that, The push rod status acquisition and zero-position calibration module includes a position acquisition and standardization submodule and a zero-position dead zone calibration and event output submodule; The position acquisition and standardization submodule is used to continuously acquire the displacement information of the push rod and convert the acquired raw displacement into a standardized position value that can be used for subsequent control. The position acquisition and standardization submodule obtains the original position sample value of the push rod and performs normalization processing according to the mechanical stroke range of the push rod to obtain the standardized position value P(t). The position acquisition and standardization submodule outputs the position sequence arranged in chronological order and adds sampling time sequence information to each sample value so that subsequent modules can calculate dynamic characteristics such as displacement change, velocity, acceleration and dwell time. The location acquisition and standardization submodule performs basic stabilization processing on the raw sampled values ​​to suppress the impact of instantaneous spike noise or abnormal jumps on subsequent control. The final output of the location acquisition and standardization submodule includes at least: the original location value, the standardized location value P(t), and the corresponding sampling timing information; The zero-position dead zone calibration and event output submodule is used to perform dead zone calibration on minute jitters near the zero position of the push rod, and outputs control event information related to the zero position based on the standardized position sequence. The zero-position dead zone calibration and event output submodule establishes a zero-position dead zone around the push rod zero position, the boundary of which is determined based on a preset threshold, equipment calibration results, or historical fluctuation statistics. The zero-position dead zone is set using a fixed threshold method or corrected using an adaptive method based on historical fluctuation statistics. When the standardized position value P(t) falls within the zero-position dead zone range, the zero-position dead zone calibration and event output submodule outputs a zero-position locked state. When the standardized position value P(t) moves from one side of the zero-position dead zone... When crossing to the other side, the zero-position dead zone calibration and event output submodule outputs a zero-crossing event signal; when the standardized position value P(t) remains in a low-change state near the zero position or the current position and continues for more than a preset duration, the zero-position dead zone calibration and event output submodule accumulates dwell time information; to avoid frequent entry and exit jitter near the zero-position boundary, the zero-position dead zone calibration and event output submodule sets an entry threshold and an exit threshold at the zero-position dead zone boundary, so that the establishment and release of the zero-position locking state have hysteresis characteristics; the final output of the zero-position dead zone calibration and event output submodule includes at least: zero-position locking state, zero-crossing event, dwell time value, and timing information corresponding to the event.

4. The system according to claim 3, characterized in that, The operation state determination and state switching module includes a dynamic feature extraction submodule, a state determination submodule, and a hysteresis switching and state output submodule. The dynamic feature extraction submodule is used to receive standardized position value P(t), zero-position locking status, zero-crossing event and dwell time information, and extract dynamic features for operation state determination based on continuous position sequence; The dynamic feature extraction submodule maintains a position sequence or sliding time window updated in chronological order, and calculates at least one or more of the following dynamic features: displacement change, velocity, acceleration or velocity change trend, dwell time, position fluctuation amplitude, and zero-position related state information. To improve the stability of dynamic features, the dynamic feature extraction submodule smooths the velocity and acceleration to suppress interference from mechanical micro-vibrations, sampling noise, or short-term abnormal jumps. The dynamic feature extraction submodule outputs a set of feature parameters for state determination, which can be used for subsequent state determination and state switching. The state determination submodule is used to determine the current push rod operation state based on the feature parameters output by the dynamic feature extraction submodule, and classify it into one of the following: rapid crossing phase, approach positioning phase, and stable dwell phase. The hysteresis switching and state output submodule is used to stabilize the state determination result and output the final effective control state. The hysteresis switching and state output submodule sets entry and exit thresholds between adjacent states to give the state switching hysteresis characteristics, avoiding frequent state jumps due to short-term fluctuations near the threshold boundaries. The hysteresis switching and state output submodule combines zero-position locking state and zero-crossing events to constrain state switching. When the system has entered a stable dwell phase, if a new zero-crossing event is detected, it exits the stable dwell phase and switches to the approach positioning phase or the rapid crossing phase. The hysteresis switching and state output submodule finally outputs the current control state S_ctrl and the corresponding state switching event information for subsequent playback timeline target generation, budget allocation, path selection, and load reduction recovery control.

5. The system according to claim 4, characterized in that, The playback timeline target generation module includes a timeline parameter generation submodule and a backfill trigger flag generation submodule; The time axis parameter generation submodule is used to receive the standardized position value P(t), the zero-position locking state and the zero-crossing event, and the current control state S_ctrl, and generate the target playback time axis parameters based on the above inputs; the target playback time axis parameters include at least the playback direction, the time scaling ratio, and the advance rate of the current playback position on the original video time axis; The time axis parameter generation submodule generates the playback direction and time scaling ratio based on the magnitude and sign of the standardized position value P(t), and sets a freeze zone near the zero position; When the standardized position value P(t) falls within the frozen zone, the time axis parameter generation submodule maintains the current valid time axis parameters, or smoothly transitions to the normal playback state according to preset rules, in order to avoid frequent changes in playback direction or advance rate due to the slight oscillation of the push rod near the zero position; the time scaling ratio and playback position advance rate are subject to upper and lower limit constraints to limit time axis abrupt changes or excessively slow advance caused by abnormal input; the time axis parameter generation submodule outputs the current playback direction, time scaling ratio, playback position advance rate, and corresponding timing information; The backfill trigger flag generation submodule combines the current control state S_ctrl with the target playback time axis parameters to determine whether the system meets the basic conditions for entering the high-quality local backfill stage, and outputs a backfill trigger flag when the conditions are met. When the system is in a stable dwell phase and the target playback time axis parameters remain stable for several consecutive sampling periods, the backfill trigger flag generation submodule outputs a high-quality backfill trigger flag. The time axis stability includes one or more of the following conditions: the playback direction has not changed, the change in the time scaling ratio is less than a preset threshold, and the current playback position is not in the unstable stage after a zero-crossing switch. When the system is detected to have exited the stable dwell phase or a new zero-crossing event occurs, the backfill trigger flag generation submodule cancels the backfill trigger flag. The backfill trigger flag generation submodule outputs the backfill trigger flag and the corresponding trigger or cancellation timing information for subsequent budget allocation and path decision-making.

6. The system according to claim 5, characterized in that, The temporal reconstruction budget allocation module includes a preview budget generation submodule, a backfill budget generation submodule, and a backfill window confirmation submodule; The preview budget generation submodule generates a preview budget B_pre that meets real-time output requirements for the preview path based on the current target playback timeline parameters and the current control state S_ctrl. The preview budget B_pre constrains the temporal reconstruction resource investment of the preview path within the current time period to prioritize the continuity of the display timeline and the low-latency following of the push bar control. The preview budget generation submodule combines the playback direction, time scaling ratio, playback position advance rate, output frame period, display cutoff constraints, and current buffer state to generate a budget set for the preview path. The budget set includes at least one or more of the following: upper limit of the number of candidate intermediate frames, generation time limit, processing complexity level, processing resolution level, and total processing allowance. When the buffer fill rate is low or the risk of display cutoff increases, the preview budget generation submodule prioritizes reducing the upper limit of the number of candidate intermediate frames, the processing complexity level, or the processing resolution level. When the buffer state is stable and the display deadline constraint is met, the preview budget generation submodule allows for an appropriate increase in preview generation capability within the budget range. The backfill budget generation submodule generates a backfill budget B_ref for local high-quality temporal reconstruction of the backfill path when the backfill trigger condition is met. When the backfill trigger flag is valid and the system is in a stable dwell phase or meets the stable dwell phase maintenance condition, the backfill budget generation submodule generates a backfill budget B_ref. The backfill budget B_ref includes at least one or more of the following: the number of candidate frames allowed for high-quality reconstruction, the processing complexity level, the processing resolution level, the processing time limit for a single backfill task, and the number of parallel backfill tasks. The establishment and improvement of the backfill budget are based on the premise that the target playback timeline remains stable, and are simultaneously constrained by the system's remaining processing capacity, buffer status, and recent backfill generation efficiency. When the system's processing capacity is insufficient, the risk of display cutoff increases, or recent backfill results are difficult to take effect in a timely manner, the backfill budget generation submodule reduces the backfill budget to avoid interference from high-quality backfill to real-time output. The backfill window determination submodule is used to determine the time window range, i.e., the local backfill window W_ref, based on the generated backfill budget B_ref. The backfill window determination submodule uses the current playback position as a reference, combined with the current playback position advance rate, the current control state S_ctrl, and the backfill budget B_ref, to determine the start and end positions of the local backfill window W_ref. The local backfill window W_ref only covers the interval containing candidate intermediate frames that have not yet reached the display deadline. The local backfill window is determined using one or a combination of the following methods: a symmetrical window extending forward and backward centered on the current playback position; a forward priority window facing the subsequent display interval; a backward compensation window covering candidate frames that have not yet been output and are to be replaced; or an asymmetric adaptive window that dynamically adjusts its boundaries according to the timeline advance direction, display queue structure, and remaining time limit. When the system just enters a stable dwell phase, an initial window length is used; when the stable dwell state continues and the backfill generation efficiency is high, the window is gradually expanded; when the system load increases or the buffer fill rate decreases, the window length is shortened accordingly.

7. The system according to claim 6, characterized in that, The dual-path strategy decision module includes a path selection submodule, a cutoff judgment display submodule, and a safety net switching submodule. The path selection submodule is used to determine the current frame interpolation path organization method adopted by the system based on the current control state S_ctrl, target playback timeline parameters, preview budget B_pre, backfill budget B_ref, and local backfill window W_ref. When the system is in the fast-crossing phase, the path selection submodule prioritizes the preview path to ensure fast timeline tracking and low-latency output. When the system is in the approaching positioning phase, the path selection submodule maintains the preview path as the dominant path and establishes conditions for enabling the backfill path. When the system is in the stable dwell phase and the backfill trigger flag is valid, the path selection submodule allows enabling the backfill path to perform high-quality backfilling of candidate intermediate frames within the local backfill window, while maintaining the preview path's responsibility for timely output. The path selection submodule selects a path organization method involving the backup path based on the current display cutoff risk and system processing capacity. The display cutoff judgment submodule is used to perform display cutoff reachability judgment on the current target to be output and the target to be rebuilt in the local backfill window. This is to determine whether the candidate version can still be generated and used for output before the display cutoff time of the corresponding display timestamp, and whether the alternative version can still be generated before the display cutoff time of the corresponding display timestamp and complete the replacement before the corresponding buffer entry enters the output lock boundary and thus take effect. The display cutoff judgment can be performed at the frame-by-frame granularity or the task-by-task granularity, and comprehensively considers factors such as the target display timestamp, the remaining processing time, the current task queue backlog, the processing load of the corresponding path, and the output lock boundary. For the candidate intermediate frames to be generated in the preview path, when the judgment result shows that it is difficult to meet the on-time display requirements under the current budget, the display cutoff judgment submodule triggers a reduction in preview processing intensity or switches to the backup path. For the replacement frame version to be generated in the backfill path, if the judgment result indicates that it is not expected to be generated before the corresponding display deadline, or is not expected to be replaced before the corresponding buffer entry enters the output locking boundary, the display deadline judgment submodule stops investing high-complexity backfill resources into the target and keeps the corresponding candidate version as a valid output version. Within the local backfill window, the display deadline judgment submodule allows reachability judgment to be performed on different target frames separately, thereby retaining only backfill tasks that still have practical replacement significance. The backup switching submodule is used to trigger the backup path to take over part or all of the output process when the current path organization cannot meet the display cutoff constraint, or when the system processing capacity is insufficient and the buffer fill rate continues to decline. The backup path adopts one or a combination of the following low-load output methods: reusing existing candidate frames or adjacent output frames, performing low-complexity simplified interpolation, directly outputting the original frame, or adopting other simplified reconstruction methods that can maintain the continuity of the current display timeline. After triggering the backup path, the backup switching submodule keeps the current target playback timeline parameters unchanged and prioritizes output continuity and display timeliness. To avoid the system frequently switching back and forth between the preview path and the backup path near the high load boundary, the backup switching submodule sets a minimum hold period after the backup path is triggered. When the display cutoff risk decreases, the buffer state is restored, and the preview path regains stable output capability, the backup path is then exited and the original path organization is restored.

8. The system according to claim 7, characterized in that, The two-stage frame interpolation execution module includes a candidate intermediate frame generation submodule, a local backfill triggering submodule, a replacement frame generation submodule, and a display timestamp binding submodule. The candidate intermediate frame generation submodule is used to perform low-latency frame interpolation on the current frame pair to be output based on the preview budget B_pre, and generate candidate intermediate frames that meet the requirements of real-time display. The candidate intermediate frame generation submodule combines the target playback timeline parameters and the current display timeline to perform temporal domain reconstruction on the original frame pair or reference frame pair corresponding to the time to be displayed, and outputs candidate intermediate frames. The candidate intermediate frame generation submodule determines the generation order of candidate intermediate frames based on the order of the target display timestamps or the urgency of the display, in order to prioritize reducing the risk of display deadline default; the generated candidate intermediate frames are used to meet the current continuous output requirements of the timeline and serve as the base version for subsequent high-quality backfilling. The local backfill trigger submodule is used to filter the set of target frames for high-quality backfilling from the local backfill window W_ref when the conditions for high-quality backfilling are met. When the backfill trigger flag is valid, the system is in a stable dwell phase, and the local backfill window W_ref is determined, the local backfill trigger submodule selects candidate intermediate frames that have generated candidate versions, have not yet been actually output, and have processing margin before the corresponding display deadline as backfillable targets. The local backfill trigger submodule comprehensively considers the target display timestamp, remaining processing time, current task queue backlog, and backfill budget B_ref to sort the backfillable targets and prioritizes the processing of target frames with actual replacement significance. When a new zero-crossing event is detected, the system exits the stable dwell phase, the display deadline risk increases, or a higher-level load reduction condition occurs, the local backfill trigger submodule stops expanding the backfill target set and cancels low-priority backfilling tasks that have not yet started execution. The replacement frame generation submodule is used to perform high-quality frame interpolation on the selected backfillable targets and generate replacement frame versions corresponding to the corresponding candidate intermediate frames. Under the constraint of backfill budget B_ref, the replacement frame generation submodule performs high-quality temporal reconstruction on the reference frame pair corresponding to the backfillable target to obtain a replacement frame with the same target temporal position as the original candidate intermediate frame. When generating replacement frames, the replacement frame generation submodule adopts a higher level of processing complexity, processing resolution, or temporal consistency constraint than the preview path. The generation of the replacement frames follows the principle of reconstruction at the same time position, that is, the replacement frame and its corresponding candidate intermediate frame are reconstructed for the same target display time. When it is determined that the current fillable target cannot be generated before the corresponding display deadline, or even if it is generated, it cannot be replaced before the corresponding buffer entry enters the output locking boundary and therefore no longer has any practical replacement meaning, the replacement frame generation submodule stops the replacement frame generation task corresponding to the current target to avoid ineffective occupation of high-quality processing resources. The timestamp binding submodule is used to bind candidate intermediate frames and alternative frame versions to their corresponding target display timestamps, and output related information for subsequent versioned buffer management. The timestamp binding submodule writes at least one or more of the following information for each candidate intermediate frame and alternative frame: display timestamp, target time domain location, version identifier, source path identifier, validity status, and corresponding display deadline. For the replacement frame version, the display timestamp binding submodule maintains the same display timestamp and the same target time domain position as the candidate intermediate frame being replaced, so that the subsequent versioning buffer module can perform in-place replacement without changing the playback timeline.

9. The system according to claim 8, characterized in that, The versioned buffer and seamless replacement output module includes a versioned buffer establishment submodule, a version replacement determination submodule, and a valid version output submodule. The versioned buffer establishment submodule is used to uniformly manage the buffers of candidate intermediate frames and replacement frame versions, and to establish a multi-version organizational structure oriented towards the same display time. The versioned buffer establishment submodule creates buffer entries based on display timestamps, and manages different frame versions corresponding to that display timestamp within each buffer entry. Only candidate versions and replacement versions with the same display timestamp and the same target time domain position are allowed to enter the same buffer entry, ensuring that subsequent replacements occur at the same target display time without changing the current playback timeline. Each buffer entry includes at least one or more of the following information: candidate version slot, replacement version slot, currently valid output version identifier, version priority, version validity status, and output lock status. For historical entries that have been output or have lost their display significance, the versioned buffer establishment submodule releases them to reclaim buffer resources. The version replacement determination submodule is used to determine whether the current replacement version is qualified to replace the corresponding candidate version, and updates the valid output version in the buffer entry when the condition is met; A replacement version is eligible to be replaced if it meets the following conditions: it has the same display timestamp and the same target time domain location as the original candidate version; and it was generated before the corresponding display deadline. The corresponding buffer entry has not yet entered the output locking boundary; the output locking boundary is set to the display deadline, or set to a reserved safety boundary earlier than the display deadline, to avoid instability caused by version switching near the actual output time; when the same display timestamp corresponds to multiple replaceable versions, the version replacement determination submodule determines the final valid output version based on version priority, generation completion time, or current valid state; when the replacement version is not generated before the deadline, or the corresponding entry has entered the output locking boundary, the candidate version continues to be used as the valid output version; The valid version output submodule is used to complete the actual display output according to the current valid output version of each buffer entry in the versioned buffer, in the order of display timestamps. Before the current display timestamp reaches the output time, the valid version output submodule reads the current valid output version from the corresponding buffer entry and sends it to the display output link. When the buffer entry has completed the valid version update before entering the output locking boundary, the updated replacement version is output; otherwise, the original candidate version is output. After the output is completed, for buffer entries that are no longer involved in subsequent display, the valid version output submodule notifies the versioned buffer to release the corresponding resources.

10. The system according to claim 9, characterized in that, The constraint closed-loop and graded load reduction module includes a constraint index monitoring submodule, a graded load reduction submodule, and a load recovery submodule. The constraint indicator monitoring submodule is used to continuously monitor key constraint indicators during system operation and provide status basis for subsequent graded load reduction and recovery control. The key constraint indicators include at least one or more of the following: display deadline default risk or default rate, processing load of preview path and backfill path, buffer fill rate or fill fluctuation, task queue backlog, backfill result generation efficiency, stability of current control state S_ctrl, and version switching conflict risk near the output lock boundary. The constraint indicator monitoring submodule does not limit its judgment of the above constraint indicators to instantaneous measurement values, but combines time window statistical results, trend judgment, or segmented threshold rules for comprehensive analysis to avoid over-response caused by short-term peak loads, occasional delays, or instantaneous state fluctuations. The constraint indicator monitoring submodule outputs the current system constraint level or operating status assessment results for subsequent load reduction and recovery control. The tiered load reduction submodule is used to implement tiered load reduction control based on the current control state S_ctrl and the system's operational risk level when constraint indicators exceed preset ranges. The tiered load reduction submodule prioritizes reducing backfill path resources, then shrinking preview path capabilities, and finally having the safety path take over partial or all output. At lower load reduction levels, the tiered load reduction submodule limits the growth of the backfill budget B_ref, shortens the local backfill window W_ref, stops adding new backfill tasks, or cancels low-priority backfill tasks that have not yet started. At higher load reduction levels, the tiered load reduction submodule further reduces the preview budget B_p. The corresponding candidate frame number limit, processing complexity level, or processing resolution level; when the display cutoff risk continues to increase, the buffer fill rate continues to decrease, or the task backlog cannot be alleviated, the graded load reduction submodule triggers the backup path to take over partial or all output, and locks the system to the minimum runnable configuration when necessary, so as to prioritize the continuous output and display timeliness of the current playback timeline; the load reduction level is executed differently according to whether it is currently in the fast crossing phase, the approach positioning phase, or the stable dwell phase, wherein in the fast crossing phase, the low latency following capability is prioritized to be protected, and in the stable dwell phase, the local high-quality backfill capability is prioritized to be compressed; The load recovery submodule is used to gradually restore the executed load reduction strategy when the system constraint indicators recover to the preset safety range and the recovery conditions are continuously met; The recovery process can be judged by combining the stability of the buffer state, the trend of the change of the display deadline default risk, the continuous stability of the current control state S_ctrl, and the efficiency of the recent backfill results. Among them, the backfill efficiency is one of the important thresholds for reopening the high-quality backfill capability. During the recovery process, if the display deadline risk increases again, the buffer fill rate decreases, the task backlog increases, or the control state switches rapidly again, further recovery will stop or revert to the previous stable recovery level to avoid the system repeatedly swinging between deload and recovery.

Images