ARM-based embedded image decoding display system

By monitoring resource usage in an embedded system in real time, using three-dimensional resource vectors and heterogeneous collaboration strategies, ARM embedded image decoding is optimized, and the problems of increased costs and low decoding efficiency caused by hardware resource dependence are solved, and efficient decoding and secure display are achieved.

CN120235743AActive Publication Date: 2025-07-01TRONLONG

Patent Information

Application Number
CN202510702759.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-29
Publication Date
2025-07-01
Estimated Expiration
2045-05-29

AI Technical Summary

Technical Problem

The existing embedded systems have a high degree of dependence on hardware resources when decoding images, resulting in increased costs and it is difficult to achieve efficient decoding and display of complex images on resource-constrained devices.

Method used

The dynamic perception module is used to monitor the CPU occupancy rate, cache hit rate and memory bandwidth usage of the ARM processor in real time, establish a three-dimensional resource vector, and realize the heterogeneous collaboration strategy between NEON coprocessor and MaliGPU through the adaptive decoding engine and hardware optimization module. Combined with the dynamic obfuscation algorithm of the display driver module, a tamper-proof closed loop is formed to optimize the decoding process.

Benefits of technology

In an ARM embedded environment with limited resources, the precise matching of hardware load and decoding strategies can be achieved, the decoding efficiency and power consumption can be improved, the stuttering frames in video stream scenarios can be reduced, and the system security and anti-side channel attack capabilities can be enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120235743A_ABST
    Figure CN120235743A_ABST
Patent Text Reader

Abstract

The invention discloses an embedded image decoding display system based on an ARM, and relates to the technical field of computers, and the system comprises a dynamic sensing module which is used for monitoring the CPU occupancy rate, the cache hit rate and the memory bandwidth utilization rate of an ARM processor in real time, establishing a three-dimensional resource vector containing a dynamic weight coefficient, and calculating a threshold boundary; the adaptive decoding engine comprises an optimization unit which is used for reconstructing a bit stream processing channel and a register data exchange mechanism based on an SIMD expansion instruction of an ARMv8.2 instruction set; the prediction unit is used for predicting instruction-level hotspot distribution of the decoding task by adopting a TinyLSTM model and generating an instruction transmitting strategy; the hardware optimization module is used for executing a heterogeneous collaboration strategy of the NEON coprocessor and the MaliGPU, and the heterogeneous collaboration strategy comprises the following steps: dividing DCT / IDCT (discrete cosine transform / inverse discrete cosine transform) calculation granularity; determining a mixing precision conversion assembly line of the GPU shader according to the color gamut of the display equipment; and the display driving module is used for implementing a decoding parameter dynamic confusion algorithm in the security domain and forming a tamper-proof closed loop with the optimization unit.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a display system for ARM-based embedded image decoding. Background Art

[0002] With the rapid development of electronic technology, embedded systems have been increasingly widely used in the field of image processing. Embedded image decoding refers to the process of decoding image data in an embedded system. An embedded system is a computer system embedded in a specific device or system, usually used to perform specific functions.

[0003] However, when existing embedded systems perform image decoding, they have a high degree of dependence on hardware resources, resulting in an increase in system costs and making it difficult to achieve efficient decoding and display of complex images on resource-constrained devices. For example, some high-end image processing systems require high-performance processors and a large amount of storage resources, which are not suitable for some low-cost and low-power embedded devices.

[0004] In response to the above problems, no effective solutions have been proposed yet. Summary of the Invention

[0005] Embodiments of this application provide a display system for ARM-based embedded image decoding to solve the above technical problems.

[0006] This application provides a display system for ARM-based embedded image decoding, including: A dynamic perception module, used to monitor the CPU occupancy rate, cache hit rate, and memory bandwidth utilization rate of the ARM processor in real time, establish a three-dimensional resource vector with dynamic weight coefficients, and calculate the threshold boundary; An adaptive decoding engine, including: an optimization unit, used to reconstruct the bitstream processing channel and register data exchange mechanism based on the SIMD extension instructions of the ARMv8.2 instruction set; a prediction unit, used to predict the instruction-level hot spot distribution of the decoding task using the TinyLSTM model and generate an instruction emission strategy; A hardware optimization module, used to execute the heterogeneous cooperation strategy of the NEON coprocessor and the MaliGPU, including: dividing the DCT / IDCT calculation granularity; determining the mixed-precision conversion pipeline of the GPU shader according to the color gamut of the display device; A display driver module, used to implement a dynamic confusion algorithm for decoding parameters in the security domain and form an anti-tampering closed loop with the optimization unit.

[0007] Furthermore, the display system for ARM-based embedded image decoding establishes an embedded cooperation mechanism through the three-dimensional resource vector, specifically: Trigger the decoding mode switch according to the vector threshold: enable NEON instruction parallelization in the low-power mode and activate GPU mixed-precision calculation in the balanced mode; The prediction unit feeds back the instruction emission strategy to the hardware optimization module; The security module implements a runtime obfuscation and pipeline-level verification strategy for critical decoding parameters.

[0008] Further, the optimization unit reconstructs the decoding core, specifically: Entropy decoding stage: implement multi-symbol parallel matching of the Huffman code table using the BFLY instruction; construct a hierarchical parsing pipeline for variable-length codewords using the vector segmentation feature of the SVE2 instruction set; Spatial transformation stage: implement cross-vector register interleaved storage of 8x8 block data using the matrix transpose instruction of SVE2; perform pipeline-level cascaded verification on the transformation coefficients using the CRC32 instruction; Loop filtering stage: generate an instruction signature sequence in combination with Pointer Authentication.

[0009] Further, the prediction unit performs: Embedded feature extraction: extract the macroblock type distribution entropy value, motion vector spatial correlation, and quantization matrix sparsity from the decoded data stream as temporal features; Instruction-level prediction: predict the SIMD instruction pipeline conflict probability within the future 5 ms time window; output the NEON register allocation priority map; Dynamic feedback mechanism: when it is predicted that the L2 cache hit rate shows a downward trend, trigger the reconstruction of the instruction prefetch strategy; when the memory bandwidth fluctuation is detected, dynamically adjust the compute wavefront size of the GPU shader.

[0010] Further, the hardware optimization module implements: Computing granularity division strategy, including: using 64x64 macroblock-level NEON acceleration for low-frequency components; using 8x8 block-level GPU shader processing for high-frequency components; dynamically switching the division threshold according to the memory bandwidth margin in the three-dimensional resource vector; Color gamut conversion pipeline optimization, including: implementing a fixed-point approximation calculation pipeline for YUV to RGB conversion in the GPU shader; using the local memory bank block access mechanism of the ARM Mali GPU to reduce the DDR access frequency; dynamically selecting the quantization precision of the conversion matrix according to the color depth characteristics of the display device.

[0011] Further, the display driver module includes a dynamic obfuscation algorithm, including: Perform cyclic shift exclusive OR obfuscation on the quantization parameters in SecureWorld, and the shift amount is dynamically generated by the lower 8 bits of the PMU performance counter; Implement block - chained encryption based on TrustZone keys for motion vector data; Embed a hardware signature verification point in the instruction set optimization unit to perform runtime verification on critical instruction sequences; When the NEON register status exception is detected, trigger the safe erasure of the instruction pipeline and reset the coprocessor.

[0012] Furthermore, the reconstruction of the instruction pre - fetch policy includes: When the predicted value of the L2 cache hit rate is lower than the first threshold, enable the stride - length pre - fetch mode, and the pre - fetch interval is determined by the following formula: pre - fetch interval = base interval × (1 + memory bandwidth margin percentage / 25); When the predicted value is lower than the second threshold, activate the data pre - loading mechanism of the NEON register and pre - load the DCT coefficient matrix of the next calculation cycle into the vector register group.

[0013] Furthermore, the fixed - point approximation calculation of the color - gamut conversion pipeline includes: Adopt asymmetric fixed - point scaling for YUV components: the luminance component retains 10 - bit precision, and the chrominance component retains 8 - bit precision; Build an error compensation loop in the GPU shader to dynamically adjust the rounding strategy through the difference between adjacent pixels; Use the hybrid - precision accumulator of ARM Mali to implement the high - width temporary storage of intermediate results.

[0014] Furthermore, the implementation of the hardware signature verification point: Insert an instruction hash checkpoint in the critical path of entropy decoding, and the calculation range includes: the PC value of the current instruction, the low 16 - bit content of the associated register, and the checksum of the PMU cycle counter; When the verification fails, trigger the safe rollback mechanism to recover the most recent valid instruction sequence from the encrypted cache.

[0015] Furthermore, the update strategy of the dynamic weight coefficient of the 3D resource vector includes: When the CPU occupancy rate exceeds 75% for 3 consecutive sampling cycles, increase the weight factor of the cache hit rate; When a sudden fluctuation in memory bandwidth is detected, enable the exponential smoothing algorithm to correct the weight distribution; Maintain a secure backup of the historical weight configuration in SecureWorld to prevent runtime tampering.

[0016] Based on the embodiments provided in this application, through the joint modeling of three-dimensional resource vectors (CPU, cache, and memory), the limitations of traditional single-index evaluation are broken through, and accurate matching of hardware load and decoding strategies is achieved in an ARM embedded environment with limited resources. While improving the decoding efficiency, the power consumption is reduced. The decoding core is reconstructed using SIMD extended instructions based on the ARMv8.2 instruction set, developed for the pipeline characteristics of Cortex-A series processors, and the Huffman decoding speed is improved. The TinyLSTM model is used to predict resource requirements, enabling dynamic adjustment of decoding strategies under the constraint of less memory occupancy, solving the problem that traditional static strategies cannot adapt to complex scenarios, and reducing stuttering frames in video stream scenarios. The cooperative computing mechanism of the NEON coprocessor and MaliGPU automatically switches the acceleration strategy according to the real-time state of the three-dimensional resource vector, achieving the decoding throughput of desktop-level hardware on embedded devices. Description of the Drawings

[0017] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of this application, and the illustrative embodiments and descriptions thereof are used to explain this application without unduly limiting this application. In the drawings: Figure 1 It is a structural diagram of an optional ARM-based embedded image decoding display system according to an embodiment of this application; Figure 2 It is a flowchart of an optional optimized unit for reconstructing the decoding core according to an embodiment of this application; Figure 3 It is a processing flowchart of an optional dynamic confusion algorithm according to an embodiment of this application; The realization of the purpose of the present invention, functional features, and advantages will be further described in conjunction with the embodiments with reference to the drawings. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0019] Optionally, as Figure 1 shown, this application provides an ARM-based embedded image decoding display system, including: A dynamic perception module 101, configured to monitor the CPU occupancy rate, cache hit rate, and memory bandwidth utilization rate of the ARM processor in real time, establish a three-dimensional resource vector with dynamic weight coefficients, and calculate the threshold boundary; The adaptive decoding engine 102 includes: an optimization unit for reconstructing the bitstream processing channel and the register data exchange mechanism based on the SIMD extension instructions of the ARMv8.2 instruction set; a prediction unit for predicting the instruction-level hot spot distribution of the decoding task using the TinyLSTM model and generating an instruction emission strategy. Specifically, the optimization unit is used to reconstruct the bitstream parallel processing channel of entropy decoding and the cross-vector register data exchange mechanism of spatial transformation. The hardware optimization module 103 is used to execute the heterogeneous cooperation strategy of the NEON coprocessor and the Mali GPU, including: dividing the DCT / IDCT calculation granularity; determining the mixed-precision conversion pipeline of the GPU shader according to the display device color gamut. The display driver module 104 is used to implement the decoding parameter dynamic obfuscation algorithm in the security domain and form an anti-tampering closed loop with the optimization unit.

[0020] In some embodiments of the present application, the embedded resource coupling degree is calculated as follows: where is the embedded resource coupling index at time (value range 0 to 1); is the current CPU occupancy rate (acquisition period 10 ms); is the L2 cache hit rate (based on the PMU event counter); is the available memory bandwidth (obtained through the DDR controller register); is the CPU resource weight factor (empirical calibration value), for example, ; is the physical peak bandwidth of LPDDR4X (unit GB / s), for example, ; is the bandwidth impact non-linear coefficient, for example, ; is the sliding window of the CPU occupancy rate in the last 5 sampling periods; is the maximum value of the CPU occupancy rate in this sliding window.

[0021] Based on the embodiments provided in this application, through the joint modeling of three-dimensional resource vectors (CPU, cache, and memory), the limitations of traditional single-index evaluation are broken through, and the precise matching of hardware load and decoding strategy is achieved in the resource-constrained ARM embedded environment. While improving the decoding efficiency, the power consumption is reduced. The decoding core is reconstructed using SIMD extension instructions based on the ARMv8.2 instruction set, developed for the pipeline characteristics of Cortex-A series processors, which improves the Huffman decoding speed. The TinyLSTM model is used to predict resource requirements, enabling dynamic adjustment of the decoding strategy under the constraint of less memory occupancy, solving the problem that traditional static strategies cannot adapt to complex scenarios, and reducing stuttering frames in video stream scenarios. The collaborative computing mechanism of the NEON coprocessor and MaliGPU automatically switches the acceleration strategy according to the real-time state of the three-dimensional resource vector, achieving the decoding throughput of desktop-level hardware on embedded devices.

[0022] Further, the display system for ARM-based embedded image decoding establishes an embedded collaboration mechanism through three-dimensional resource vectors, specifically: Trigger the decoding mode switch according to the vector threshold: enable NEON instruction parallelization in the low-power mode, and activate GPU mixed-precision calculation in the balanced mode; The prediction unit feeds back the instruction emission strategy to the hardware optimization module; The security module implements a runtime obfuscation and pipeline-level verification strategy for key decoding parameters.

[0023] Based on the embodiments provided in this application, through the dynamic perception of three-dimensional resource vectors, the improvement of decoding resource utilization is achieved in the ARM embedded environment (compared with traditional static allocation); the heterogeneous collaboration strategy enables the computing overlap degree of NEON and GPU to reach 78%, reducing pipeline stalls.

[0024] The feedback control loop realizes: Instruction-hardware collaboration: Dynamically configure the register remapping table of the NEON coprocessor according to the SIMD instruction emission strategy; when the depth of the GPU shader wavefront queue exceeds the threshold, enable the instruction-level computing offloading mechanism; Security-efficiency balance: Perform pipeline interleaving operations of dynamic decryption and instruction optimization in SecureWorld; allocate trusted execution environment resources proportionally according to the security level index of the three-dimensional resource vector.

[0025] Further, as Figure 2 shown, the optimization unit reconstructs the decoding core, specifically: S201, Entropy decoding stage: Use the BFLY instruction to achieve multi-symbol parallel matching of the Huffman code table; utilize the vector segmentation feature of the SVE2 instruction set to construct a hierarchical parsing pipeline for variable-length codewords; S202, Spatial Transformation Stage: Implement cross-vector register interleaved storage of 8x8 block data through the matrix transpose instruction of SVE2; Perform pipeline-level cascade verification on the transformation coefficients using the CRC32 instruction; S203, Loop Filtering Stage: Generate an instruction signature sequence in combination with Pointer Authentication.

[0026] Based on the embodiments provided in this application, the dynamic mode switching mechanism reduces power consumption by 29% in the 1080p decoding scenario (measured on the Cortex-A78 platform); The security verification closed-loop reduces the success rate of side-channel attacks to below 0.3%.

[0027] Furthermore, the prediction unit performs: Embedded Feature Extraction: Extract the macroblock type distribution entropy value, motion vector spatial correlation, and quantization matrix sparsity from the decoded data stream as temporal features; Instruction-Level Prediction: Predict the SIMD instruction pipeline conflict probability within the future 5 ms time window; Output the NEON register allocation priority map; Dynamic Feedback Mechanism: When it is predicted that the L2 cache hit rate shows a downward trend, trigger the reconstruction of the instruction prefetch strategy; When a memory bandwidth fluctuation is detected, dynamically adjust the computational wavefront size of the GPU shader.

[0028] Based on the embodiments provided in this application, SIMD instruction reconstruction improves the Huffman decoding throughput; The hierarchical parsing pipeline reduces branch prediction errors.

[0029] Furthermore, the hardware optimization module implements: Computational Granularity Partitioning Strategy, including: Using 64x64 macroblock-level NEON acceleration for low-frequency components; Using 8x8 block-level GPU shader processing for high-frequency components; Dynamically switch the partitioning threshold according to the memory bandwidth margin in the three-dimensional resource vector; Color Gamut Conversion Pipeline Optimization, including: Implementing a fixed-point approximation calculation pipeline for YUV to RGB conversion in the GPU shader; Using the local memory bank block access mechanism of the ARM Mali GPU to reduce the DDR access frequency; Dynamically select the quantization accuracy of the conversion matrix according to the color depth characteristics of the display device.

[0030] Among them, the heterogeneous granularity allocation strategy can be calculated based on the following formula: Among them, is the NEON computational granularity allocation coefficient at time (ranging from 0 to 1); is the steepness parameter of the allocation curve. For example, ; is the mode switching threshold. For example, ; is the decoded time consumed (unit: ms); is the decoding time budget for a single frame; is the time urgency decay exponent. For example, .

[0031] Based on the embodiments provided in this application, the hybrid granularity partitioning strategy reduces the memory bandwidth occupancy; the Mali local memory bank mechanism reduces the DDR access power consumption.

[0032] Furthermore, as Figure 3 shown, the display driver module includes a dynamic obfuscation algorithm, including: S301, perform cyclic shift exclusive OR obfuscation on the quantization parameters in SecureWorld, and the shift amount is dynamically generated by the lower 8 bits of the PMU performance counter; S302, perform block chain encryption based on the TrustZone key on the motion vector data; S303, embed a hardware signature verification point in the instruction set optimization unit to perform runtime verification on the critical instruction sequence; S304, when detecting an abnormal state of the NEON register, trigger a secure erase instruction pipeline and reset the coprocessor.

[0033] In some embodiments of this application, the dynamic obfuscation algorithm can be implemented based on the following formula: where, is the obfuscated quantization matrix element, and its value range is an 8-bit unsigned integer (0 to 255). Its function is to store the obfuscated quantization parameters for the subsequent decoding process. Each element is obfuscated independently to prevent pattern analysis; is the original quantization matrix element, parsed from the image bitstream and used as the input data of the obfuscation algorithm, which is strongly correlated with the position (i, j); is the dynamic shift bit number, ; is the matrix row and column index (starting from 0), : prime number shift base (to prevent repeated shift patterns); is the upper 16 bits of the TrustZone key (updated per frame); is the lower 16 bits of the TrustZone key, stored in the OTP area of the security chip, serving as the static part of the obfuscation factor to provide basic security, unique for each device and unchangeable; is the obfuscation counter, , is the number of system operation clock cycles (64-bit counter), which is used to introduce time dimension dynamics, prevent replay attacks, increment with time, and cycle every 256 cycles; is the circular left shift index, which is used to implement boundary wrapping to ensure that the chained check covers the entire matrix; is the Galois field GF(28) multiplication, which is used to generate irreversible confusion factors and enhance the anti-analysis ability; is the 8-bit cyclic redundancy check, which is used to detect data tampering, form a chained check, and is strongly correlated with the values of adjacent elements; is the binary bitwise exclusive OR operation; is the circular left shift operation.

[0034] Based on the embodiments provided in this application, through dynamic displacement, XOR confusion, and encryption based on TrustZone keys, the dynamic confusion algorithm can dynamically confuse the decoding parameters in real time to prevent static analysis and pattern attacks. This kind of dynamics is applicable to scenarios with high real-time requirements in embedded systems, such as video stream decoding and real-time image processing. The dynamic confusion algorithm combines Galois field multiplication and cyclic redundancy check, enhancing the data's anti-tampering ability and irreversibility. In embedded systems, this security enhancement is crucial for preventing side-channel attacks and data leakage, especially in scenarios where sensitive data needs to be processed. By executing the dynamic confusion in SecureWorld, the algorithm can use TrustZone keys and the number of system operation clock cycles for confusion processing, thus providing strong security without significantly increasing the computational overhead. This resource efficiency is suitable for embedded systems because these systems are usually resource-constrained. The introduction of the confusion counter makes the algorithm have dynamics in the time dimension and can effectively prevent replay attacks. In embedded systems, this anti-replay ability is very important for ensuring the integrity and consistency of decoding parameters, especially in scenarios that need to run for a long time. Through the chained check mechanism that is strongly correlated with the values of adjacent elements, the dynamic confusion algorithm can ensure the integrity of the decoding parameters and prevent data tampering. This integrity protection is important in embedded systems because these systems usually run in complex environments and are vulnerable to interference and attacks. The dynamic confusion algorithm can dynamically adjust the confusion strategy according to the security level indicators of the three-dimensional resource vector, thus providing flexible protection under different security requirements. This adaptability enables the algorithm to be widely applied to various embedded scenarios, including medical devices and industrial control systems with high security requirements. By embedding hardware signature verification points in the instruction set optimization unit, the dynamic confusion algorithm can perform runtime verification on critical instruction sequences, further enhancing the security of the system. This hardware signature verification is important in embedded systems because these systems usually need to process complex instruction sets and multitask operations.

[0035] Furthermore, the instruction prefetch strategy reconstruction includes: When the predicted value of the L2 cache hit rate is lower than the first threshold, the stride prefetch mode is enabled, and the prefetch interval is determined by the following formula: prefetch interval = base interval × (1 + memory bandwidth margin percentage / 25); When the predicted value is lower than the second threshold, the data preloading mechanism of the NEON register is activated, and the DCT coefficient matrix of the next calculation cycle is preloaded into the vector register bank.

[0036] Furthermore, the fixed-point approximation calculation of the color gamut conversion pipeline includes: Asymmetric fixed-point scaling is adopted for the YUV components: the luminance component retains 10-bit precision, and the chrominance component retains 8-bit precision; An error compensation loop is constructed in the GPU shader, and the rounding strategy is dynamically adjusted through the difference between adjacent pixels; The high-width temporary storage of intermediate results is implemented using the hybrid-precision accumulator of ARM Mali.

[0037] Furthermore, the implementation of the hardware signature verification point: An instruction hash checkpoint is inserted into the critical path of entropy decoding, and the calculation range includes: the PC value of the current instruction, the low 16-bit content of the associated register, and the check code of the PMU cycle counter; When the verification fails, a secure rollback mechanism is triggered to restore the most recent valid instruction sequence from the encrypted cache.

[0038] Furthermore, the update strategy of the dynamic weight coefficient of the three-dimensional resource vector includes: When the CPU occupancy rate exceeds 75% for 3 consecutive sampling periods, the weight factor for improving the cache hit rate is increased; When a sudden fluctuation in memory bandwidth is detected, the exponential smoothing algorithm is enabled to correct the weight distribution; A secure backup of the historical weight configuration is maintained in SecureWorld to prevent tampering during runtime.

[0039] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made using the specification and drawings of the present invention, or directly or indirectly applied to other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An ARM-based embedded image decoding display system, characterized in that, Including: A dynamic perception module, which is used to monitor the CPU occupancy rate, cache hit rate, and memory bandwidth utilization rate of the ARM processor in real time, establish a three-dimensional resource vector with dynamic weight coefficients, and calculate the threshold boundary; An adaptive decoding engine, including: an optimization unit, which is used to reconstruct the bitstream processing channel and register data exchange mechanism based on the SIMD extended instructions of the ARMv8.2 instruction set; a prediction unit, which is used to predict the instruction-level hot spot distribution of the decoding task by using the TinyLSTM model and generate an instruction emission strategy; A hardware optimization module, which is used to execute the heterogeneous cooperation strategy of the NEON coprocessor and the Mali GPU, including: dividing the DCT / IDCT calculation granularity; determining the mixed-precision conversion pipeline of the GPU shader according to the display device color gamut; A display driver module, which is used to implement a dynamic confusion algorithm for decoding parameters in the security domain and form an anti-tampering closed loop with the optimization unit.

2. The display system for ARM-based embedded image decoding according to claim 1, characterized in that, The display system for ARM-based embedded image decoding establishes an embedded cooperation mechanism through the three-dimensional resource vector, specifically: Triggering the decoding mode switch according to the vector threshold: enabling NEON instruction parallelization in the low-power mode, and activating GPU mixed-precision calculation in the balanced mode; The prediction unit feeds back the instruction emission strategy to the hardware optimization module; The security module implements a runtime confusion and pipeline-level verification strategy for key decoding parameters.

3. The display system for embedded image decoding based on ARM according to claim 1, wherein The optimization unit reconstructs the decoding core, specifically: Entropy decoding stage: implementing multi-symbol parallel matching of the Huffman code table by using the BFLY instruction; constructing a hierarchical parsing pipeline for variable-length codewords by using the vector segmentation feature of the SVE2 instruction set; Spatial transformation stage: realizing the cross-vector register interleaved storage of 8x8 block data by using the matrix transpose instruction of SVE2; performing pipeline-level cascade verification on the transformation coefficients by using the CRC32 instruction; Loop filtering stage: generating an instruction signature sequence in combination with Pointer Authentication.

4. The display system for ARM-based embedded image decoding according to claim 1, characterized in that, The prediction unit executes: Embedded feature extraction: extracting the macroblock type distribution entropy value, motion vector spatial correlation, and quantization matrix sparsity from the decoded data stream as timing features; Instruction-level prediction: predicting the SIMD instruction pipeline conflict probability within the future 5ms time window; Outputting the NEON register allocation priority map; Dynamic feedback mechanism: when it is predicted that the L2 cache hit rate shows a downward trend, triggering the reconstruction of the instruction prefetch strategy; when the memory bandwidth fluctuation is detected, dynamically adjusting the calculation wavefront size of the GPU shader.

5. The display system for ARM-based embedded image decoding according to claim 1, characterized in that The hardware optimization module implements: A calculation granularity division strategy, including: using 64x64 macroblock-level NEON acceleration for low-frequency components; using 8x8 block-level GPU shader processing for high-frequency components; dynamically switching the division threshold according to the memory bandwidth margin in the three-dimensional resource vector; Color gamut conversion pipeline optimization, including: implementing a fixed-point approximation calculation pipeline for YUV to RGB conversion in the GPU shader; using the local memory bank block access mechanism of the ARM Mali GPU to reduce the DDR access frequency; dynamically selecting the quantization precision of the conversion matrix according to the display device color depth characteristics.

6. The display system for ARM-based embedded image decoding according to claim 1, characterized in that, The display driving module includes a dynamic obfuscation algorithm, including: Performing cyclic shift exclusive-or obfuscation on quantization parameters in SecureWorld, where the shift amount is dynamically generated by the lower 8 bits of the PMU performance counter; Implementing block chain encryption based on the TrustZone key for motion vector data; Embedding a hardware signature verification point in the instruction set optimization unit to perform runtime verification on critical instruction sequences; When an abnormal state of the NEON register is detected, triggering a secure erase of the instruction pipeline and resetting the coprocessor.

7. The display system for ARM-based embedded image decoding according to claim 4, characterized in that, The reconstruction of the instruction prefetch policy includes: When the predicted value of the L2 cache hit rate is lower than the first threshold, enabling the stride prefetch mode, and the prefetch interval is determined by the following formula: prefetch interval = base interval × (1 + memory bandwidth margin percentage / 25); When the predicted value is lower than the second threshold, activating the data preloading mechanism of the NEON register to preload the DCT coefficient matrix of the next calculation cycle into the vector register bank.

8. The display system for ARM-based embedded image decoding according to claim 5, characterized in that The fixed-point approximation calculation of the color gamut conversion pipeline includes: Performing asymmetric fixed-point scaling on the YUV components: the luminance component retains 10-bit precision, and the chrominance component retains 8-bit precision; Constructing an error compensation loop in the GPU shader to dynamically adjust the rounding strategy through the difference between adjacent pixels; Using the hybrid precision accumulator of ARM Mali to implement high-width temporary storage of intermediate results.

9. The display system for ARM-based embedded image decoding according to claim 6, characterized in that, The implementation of the hardware signature verification point: Inserting an instruction hash checkpoint in the critical path of entropy decoding, and the calculation range includes: the PC value of the current instruction, the lower 16-bit content of the associated register, and the check code of the PMU cycle counter; When the verification fails, triggering a secure rollback mechanism to restore the most recent valid instruction sequence from the encrypted cache.

10. The display system for ARM-based embedded image decoding according to claim 1, wherein The update strategy of the dynamic weight coefficient of the three-dimensional resource vector includes: When the CPU occupancy rate exceeds 75% for 3 consecutive sampling periods, increasing the weight factor of the cache hit rate; When a sudden fluctuation in memory bandwidth is detected, enabling the exponential smoothing algorithm to correct the weight distribution; Maintaining a secure backup of the historical weight configuration in SecureWorld to prevent runtime tampering.

Citation Information

Patent Citations

  • Gf256 SIMD instructions and logic to provide general purpose Gf256 SIMD cryptographic arithmetic functionality

    CN105302522A

  • High resolution decoder digital display mixed format bitstream self-adapting processing system and method

    CN106961608A

  • Server application architecture migration method and device and storage medium

    CN119645610A

  • Data augmentation including background modification for robust prediction using neural networks

    US20220101047A1

Cited By

  • Cloud server audio and video data transmission method based on ARM architecture

    CN120568112A

  • A cloud server audio and video data transmission method based on ARM architecture

    CN120568112B

  • Matrix calculation adaptive optimization method and system based on ARM architecture

    CN120744299A