A color screen instrument image classification method and system based on RLE decompression

Abstract

The present application relates to the technical field of image processing, and provides a color screen instrument image classification method and system based on RLE decompression, which comprises the following steps: acquiring vehicle running state data, and performing first verification on a first identification result based on the vehicle running state data to obtain a first verification conclusion; performing second identification on a local area containing safety critical information in a data stream to obtain a second identification conclusion; if the safety critical information category indicated by the first identification result is consistent with the safety critical information categories indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority; if the safety critical information category indicated by the first identification result is inconsistent with the safety critical information categories indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed. The present application has the effect of improving the accuracy and reliability of color screen instrument image classification.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and specifically to a method and system for classifying color screen instrument images based on RLE decompression. Background Technology

[0002] In modern intelligent driving vehicles, the color instrument panel system is the core window for drivers to obtain vehicle information and a key interface for realizing intelligent driving and in-vehicle interaction. To efficiently utilize limited storage space and ensure smooth data transmission, instrument panels typically employ compression technologies such as travel length encoding (RLE) to store various image data, including navigation maps, multimedia interfaces, vehicle status displays, and various warning icons. When the system needs to present these compressed images to the driver, it must perform RLE decompression to restore the data to displayable pixel information. However, simply decompressing the images is insufficient; the system also needs to quickly and accurately identify the content of these decompressed images to determine how to perform subsequent display processing based on their category, such as overlaying routes on a map or displaying real-time vehicle speed on the instrument panel.

[0003] In existing technologies, image content recognition is typically performed by a separate classification module after RLE decompression. This sequential "decompress first, classify later" approach often leads to significant system latency when facing the increasing demands of intelligent driving for real-time and rich display content, thus affecting the driver's experience and driving safety. To improve system response speed, an improved solution has been proposed that performs image classification simultaneously during RLE decompression. This integrated decompression and classification process eliminates the intermediate step of generating a complete bitmap for classification and the corresponding memory usage, theoretically significantly reducing display latency.

[0004] However, in real-world vehicle applications, this integrated solution faces significant reliability issues. The graphical resource files for in-vehicle systems may be created by different design teams at different times or provided by different suppliers, leading to inconsistencies within the resource files themselves. For example, a low tire pressure warning icon might have a specific amber background color according to design specifications, but multiple versions might exist in the actual stored resource files. One version might have a slightly different background color value due to incorrect export settings. For integrated classification methods that directly analyze RLE data streams, since their judgment criteria may include precise matching of specific color values, such color value deviations could cause the classifier to fail to recognize it as a warning icon, resulting in missed warnings and posing a safety hazard.

[0005] A more serious problem than color deviation is data corruption. Over long-term use, in-vehicle storage devices may develop occasional bit errors due to hardware aging or electromagnetic interference. The RLE data format is highly sensitive to such errors; a single bit flip can turn a short stroke length value into an extremely large number, or a color value into a completely different color. When the decompressor processes this corrupted data, it may draw a long, incorrectly colored line on the screen. The integrated classifier, analyzing this contaminated data stream, will extract completely erroneous features, potentially misidentifying a red warning icon that should indicate "engine malfunction" as a "multimedia playback" interface, thus presenting the user with entirely incorrect information—absolutely unacceptable while driving.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] This application discloses a color screen instrument image classification method and system based on RLE decompression, aiming to solve the problems of system delay and reliability in image classification after RLE decompression in the prior art, as well as classification errors and security risks caused by inconsistent resource files or data corruption.

[0008] The technical solution of this application is as follows:

[0009] In a first aspect, this application discloses a color screen instrument image classification method based on RLE decompression, including:

[0010] During the decompression process of the RLE compressed image data stream, the data stream is initially identified based on the preset safety-critical RLE mode signature. The preset safety-critical RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety-critical display area. The signature template includes at least the key area identifier, key color category identifier, run length sequence feature, and matching threshold configuration.

[0011] When safety-critical information is identified during the initial identification process, a first identification result is output and a preset high-priority verification process is initiated. The high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety-critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on the local area containing the safety-critical information in the data stream, the second identification being based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion.

[0012] Based on the first identification result, the first verification conclusion, and the second identification conclusion, a final confirmation is made. If the category of safety critical information indicated by the first identification result is consistent with the category of safety critical information indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the category of safety critical information indicated by the first identification result is inconsistent with the category of safety critical information indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

[0013] This technical solution enables preliminary identification to be performed simultaneously during RLE decompression, effectively improving processing efficiency. Furthermore, a high-priority verification process is introduced, combining first verification with vehicle operating status data and second identification with local pixel features to perform multi-dimensional cross-verification of the preliminary identification results. This significantly enhances the accuracy and reliability of safety-critical information identification, effectively avoiding false alarms and missed alarms caused by data inconsistency or corruption. This solves the reliability problem of integrated classification schemes in existing technologies, ensuring driving safety.

[0014] Secondly, this application also discloses a color screen instrument image classification system based on RLE decompression, used to perform color screen instrument image classification based on RLE decompression, including:

[0015] The preliminary identification execution module is used to perform preliminary identification of the data stream based on the preset safety-critical RLE mode signature during the decompression process of the RLE compressed image data stream. The preset safety-critical RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety-critical display area. The signature template includes at least a key area identifier, a key color category identifier, run length sequence features, and a matching threshold configuration.

[0016] The identification and verification conclusion module is used to output a first identification result and initiate a preset high-priority verification process when safety-critical information is identified during the initial identification process. The high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety-critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on the local area containing the safety-critical information in the data stream, the second identification being based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion.

[0017] The identification result processing module is used to make a final confirmation based on the first identification result, the first verification conclusion, and the second identification conclusion. If the safety critical information category indicated by the first identification result is consistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the safety critical information category indicated by the first identification result is inconsistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

[0018] This application provides a color screen instrument panel image classification system based on RLE decompression. Through modular design, it achieves preliminary identification simultaneously during RLE decompression, and performs multiple verifications by combining external diagnostics and local image features. Finally, it performs intelligent display processing based on the verification results. This system can efficiently and accurately identify safety-critical information, effectively solving the delay and reliability problems of image classification in existing technologies, and providing a safer and more reliable instrument panel display solution for intelligent driving vehicles.

[0019] Beneficial Effects: Upon initial identification of safety-critical information, this application initiates a high-priority verification process. This process includes sending a diagnostic request to an external control unit to obtain vehicle operating status data, and performing a first verification based on this data for the initial identification result. Simultaneously, a second identification is performed on the local area containing the safety-critical information, based on the relative color relationships and structural characteristics between pixels within the local area. This multi-dimensional verification mechanism effectively solves the problem in existing technologies where classifiers cannot accurately identify information due to inconsistent resource files (such as color deviations) or data corruption (such as bit errors). By combining external diagnostic information about the vehicle's operating status with the inherent visual features of the image itself, this application can cross-validate the initial identification result, greatly enhancing the accuracy and reliability of the identification.

[0020] Ultimately, this application makes a final confirmation based on the first identification result, the first verification conclusion, and the second identification conclusion. If the categories of safety-critical information indicated by the three are consistent, they are displayed with the highest priority, ensuring the timely and accurate presentation of critical information. If they are inconsistent, the first identification result is not adopted and a rollback operation is performed, effectively avoiding the display of erroneous information, thereby solving the serious safety hazard in the prior art, such as misidentifying "engine malfunction" as "multimedia playback".

[0021] In summary, this application overcomes the technical challenges of image classification delay, resource file inconsistency, and data corruption in existing technologies by integrating decompression and classification, multi-dimensional verification, and intelligent decision rollback mechanism. It significantly improves the real-time performance, accuracy, and reliability of color screen instrument image classification, providing a safer and more efficient display information processing solution for intelligent driving vehicles. Attached Figure Description

[0022] Figure 1 This is a flowchart of a color screen instrument image classification method based on RLE decompression in one embodiment of the present invention;

[0023] Figure 2 This is a flowchart of a color screen instrument image classification method based on RLE decompression in another embodiment of the present invention;

[0024] Figure 3 This is a system block diagram of a color screen instrument image classification system based on RLE decompression according to another embodiment of the present invention;

[0025] Explanation of reference numerals in the attached figures:

[0026] 1. Color screen instrument image classification system based on RLE decompression; 11. Preliminary recognition execution module; 12. Recognition verification conclusion module; 13. Recognition result processing module. Detailed Implementation

[0027] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0028] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0029] This application proposes a color screen instrument image classification method based on RLE decompression, combined with... Figure 1 As shown, it includes:

[0030] S1, During the decompression process of the RLE compressed image data stream, the data stream is initially identified based on the preset safety key RLE mode signature; The preset safety key RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety key display area. The signature template includes at least the key area identifier, the key color category identifier, the run length sequence feature, and the matching threshold configuration.

[0031] S2, when safety-critical information is identified during the initial identification process, the first identification result is output and a preset high-priority verification process is initiated; the high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety-critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on the local area containing the safety-critical information in the data stream, the second identification being based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion;

[0032] S3. Based on the first identification result, the first verification conclusion, and the second identification conclusion, a final confirmation is made. If the safety critical information category indicated by the first identification result is consistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the safety critical information category indicated by the first identification result is inconsistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

[0033] To better understand the technical solution proposed in this application, the key terms involved will be explained first.

[0034] "RLE compressed image data stream" refers to an image data sequence compressed using run-length encoding. The basic characteristic of this data stream is that it combines and encodes consecutively repeating pixel values ​​and their corresponding repetition counts to reduce the image data size, thereby improving transmission or processing efficiency.

[0035] The "Safety-Critical RLE Pattern Signature" is a pre-defined template information used to characterize the run sequence of a specific safety-critical display area in the data stream of an RLE-compressed image. This Safety-Critical RLE Pattern Signature includes at least: a critical region identifier, used to characterize the approximate area of ​​the safety-critical information in the image; a critical color category identifier, used to define the core color category corresponding to the safety-critical information; run length sequence features, used to describe the characteristics of the RLE runs constituting the safety-critical information in terms of length distribution and arrangement order; and a matching threshold configuration, used to limit the allowable deviation range during the recognition process.

[0036] An "external control unit" refers to an electronic control unit in a vehicle that performs specific control functions, such as the engine control unit, brake control unit, and tire pressure monitoring control unit. This external control unit can output data related to the vehicle's current operating status for cross-validation by the display and recognition system.

[0037] "Vehicle operating status data" refers to data output by an external control unit that reflects the current actual operating status of the vehicle. Specifically, it may include vehicle speed, engine load, ambient temperature, tire pressure data, braking status data, etc.

[0038] Based on the aforementioned terminology, the method proposed in this application, during the decompression process of an RLE compressed image data stream, first performs preliminary identification of the data stream based on a preset security-critical RLE pattern signature. This preliminary identification step is necessary because the RLE compressed image data stream itself does not directly correspond to a complete pixel plane. If the judgment is made directly based on the subsequent complete decoding result without targeted identification, it can easily increase processing latency and hinder the timely discovery of security-critical information. By directly combining the security-critical RLE pattern signature with preliminary identification during the decompression process, target areas that may correspond to security-critical information can be discovered before the data is fully expanded, thus buying time for subsequent verification and display.

[0039] Specifically, preliminary identification can be implemented in several ways. First, during decompression, the data stream can be scanned byte-by-byte or line-by-line, and the scan results compared with a preset safety-critical RLE pattern signature. When a segment in the data stream matches the safety-critical RLE pattern signature in terms of key region identification, key color category identification, and line length sequence features within the allowed range of the matching threshold configuration, the segment is determined to correspond to safety-critical information. Second, continuous lines can be extracted during decompression to construct a simplified feature vector containing color, length, and position. This simplified feature vector is then compared with a preset feature vector in the safety-critical RLE pattern signature for similarity calculation. When the similarity reaches the matching threshold configuration requirement, safety-critical information is also preliminarily identified. The necessity of using these methods lies in the fact that RLE data retains the structural and color distribution information of local areas in its compressed state. Utilizing this information for early identification facilitates rapid screening of safety-critical information without relying entirely on the reconstruction of the entire image.

[0040] When safety-critical information is identified during the initial identification process, the system outputs the first identification result and initiates a preset high-priority verification process. Setting up a high-priority verification process is necessary because the initial identification is essentially a rapid judgment based on compressed data characteristics. If the RLE data stream has partial corruption, compression errors, color perturbations, or abnormal flow patterns, relying solely on the first identification result may still lead to misidentification. Therefore, further invoking the high-priority verification process after the first identification result is generated helps improve the reliability of safety-critical information determination.

[0041] The first identification result is used to characterize the category and corresponding location of the safety-critical information obtained from the initial identification. In the high-priority verification process, the system sends a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the first identification result. The necessity of setting up this verification step lies in the fact that the displayed content itself belongs to the image data domain, while the vehicle operating status data belongs to the vehicle control domain, and the two have different sources. If the identification result in the display domain can be supported by the status data in the control domain, then the first identification result can be externally verified, thereby reducing the risk of false alarms. For example, when the first identification result indicates "low tire pressure," the system can send a diagnostic request to the external control unit related to tire pressure monitoring to obtain the real-time tire pressure data of the corresponding wheel position; if the vehicle operating status data shows that the actual tire pressure is indeed lower than a preset threshold, then the first verification conclusion supports the first identification result. Conversely, if the vehicle operating status data does not support the identified content, it indicates that the possibility of a deviation in the first identification result has increased.

[0042] Furthermore, the high-priority verification process also includes performing a second identification on local areas containing safety-critical information in the data stream to obtain a second identification conclusion. The second identification is based on the relative color relationships and structural characteristics between pixels within the local area. The necessity of setting up the second identification lies in the fact that the first identification mainly relies on RLE compression features for rapid matching, while the second identification turns to the local image content itself, reconfirming the identification results by analyzing the relative color relationships, edge structures, icon outlines, and local morphological features within the local area. This provides supplementary judgment on safety-critical information from another dimension, avoiding misjudgments caused by solely relying on RLE travel patterns. For example, when the first identification result indicates that a certain local area corresponds to an "engine malfunction" warning icon, the second identification can further analyze the color distribution relationship, edge clarity, and specific outline structure within that local area to determine whether it conforms to preset icon characteristics. Through this local area identification method, even if there are travel anomalies or color shifts in the RLE data in a local area, the identification results can still be corrected using structural characteristics.

[0043] After completing the above verification, the system performs a final confirmation based on the first identification result, the first verification conclusion, and the second identification conclusion. The necessity of setting up this final confirmation step lies in the fact that the identification or verification conclusion of a single path may be limited by its own data source and processing method. However, by jointly judging the preliminary identification result, the external control unit verification result, and the second identification result of the local area, a comprehensive confirmation mechanism with multiple sources and multiple levels can be formed, thereby improving the robustness of the confirmation results of safety-critical information.

[0044] Specifically, when the category of safety-critical information indicated by the first identification result is consistent with the category of safety-critical information indicated by the first verification conclusion and the second identification conclusion, the system performs the highest priority display processing on the first identification result. The necessity of this setting lies in the fact that when the conclusions of compressed data domain identification, control domain verification, and local area image feature identification are consistent, it indicates that the safety-critical information has a high degree of credibility, and its display should be prioritized so that the driver can be promptly informed of the relevant safety status. Conversely, when the category of safety-critical information indicated by the first identification result is inconsistent with the first verification conclusion or the second identification conclusion, the first identification result is not adopted, and a rollback operation is performed. The necessity of setting up the rollback operation lies in the fact that if the first identification result is still directly used for display when the multi-path verification results are inconsistent, it may convey incorrect warning information to the driver, thereby affecting driving judgment. The rollback operation may include re-performing identification, displaying general warning information, or temporarily not displaying the corresponding identification result to avoid unnecessary interference caused by false alarms.

[0045] In summary, this application achieves hierarchical identification and verification of safety-critical information in RLE compressed image data streams through a processing chain of "preliminary identification based on safety-critical RLE pattern signatures, first verification based on external control units, second identification based on local region internal features, and final confirmation based on multi-result consistency." This approach retains the response speed advantage of direct identification of compressed data while improving the reliability of safety-critical information identification results through cross-verification using external control units and local region image features.

[0046] Optional, combined Figure 2 As shown, the steps for preliminary identification of the data stream based on the preset security-critical RLE pattern signature during the decompression process of the RLE compressed image data stream include:

[0047] A1. Perform rule-based aggregation on the data stream of the RLE compressed image. The rule-based aggregation process includes: at the start of rule-based aggregation, if a segment of the expected main color tone is interrupted by a segment within the window, rule-based aggregation judgment is initiated; where the window is a preset decompression scan range; the segment is the length of consecutive pixels of the same color in the RLE; and the segment is a segment of the segment that has been interrupted or split. After initiating rule-based aggregation judgment, it is determined whether the length of the interrupted segment is lower than the segment length threshold, and a segment length judgment result is obtained; it is determined whether the hue difference and saturation difference between the color of the interrupted segment and the expected main color tone in the color space are both less than the difference threshold, and a color difference judgment result is obtained; when both the segment length judgment result and the color difference judgment result are yes, the interrupted segments are merged or removed according to preset rules and reconstructed into continuous segments; where the process of merging or removing the interrupted segments according to preset rules is: when the length of the interrupted segment is lower than the preset length threshold and the color difference with the adjacent main segment is lower than the preset color threshold, the interrupted segment is merged into the adjacent main segment; otherwise, the interrupted segment is removed when generating continuous segments.

[0048] A2, based on the continuous process obtained after rule aggregation, is matched with the preset safety-critical RLE pattern signature to perform preliminary identification of the data flow.

[0049] Specifically, in the above rule aggregation process, the setting of each judgment condition and processing flow is clearly necessary, and together they constitute a robust processing mechanism for non-ideal encoding situations in RLE data streams.

[0050] The necessity of introducing a rule aggregation mechanism during RLE data stream decompression lies in the fact that RLE encoding is easily affected by quantization errors, sampling jitter, or differences in compression strategies during actual generation and transmission, causing the originally continuous main run to be split into multiple short run segments. If pattern matching is performed directly based on the original run sequence, these minor interruptions will cause the run length sequence features to shift, resulting in deviations from the matching results with the safety-critical RLE pattern signature. Therefore, by performing rule aggregation on local data during decompression, the continuity of the main run can be restored without changing the overall image structure, thus providing a more stable input basis for subsequent matching.

[0051] The necessity of setting a window lies in the fact that rule aggregation is a local correction process. If all processes are processed directly in the global scope, the computational complexity will increase significantly and cross-regional interference may be introduced. By limiting the window range, the aggregation judgment is performed only within a local data segment, which can ensure the temporal continuity and spatial consistency of the processing, while avoiding the mutual influence of features between different image regions, thereby improving processing efficiency and ensuring the stability of the results.

[0052] Introducing a run-length threshold in rule aggregation judgment is necessary because it distinguishes between "true structural details" and "short run-length segments caused by noise." Run-length is a crucial criterion for differentiation; shorter runs are more likely to originate from coding perturbations or non-critical details, while longer runs typically constitute the main structure of the image. By setting a run-length threshold, short run-length segments can be filtered out and included as candidate interruptions in subsequent processing, thus avoiding misprocessing of the main structure.

[0053] Simultaneously, introducing a color difference threshold (including hue and saturation differences) is necessary because judging solely by length cannot identify color abrupt changes. If a short segment has a significant color difference, it may belong to a true boundary or structural feature and should not be merged. By measuring the difference between the short segment and the dominant hue in the color space and setting a difference threshold, the merging conditions can be further constrained, ensuring that rule-based aggregation only applies to areas with high color continuity, thereby avoiding damage to the true image structure.

[0054] When an interrupted segment satisfies both length and color criteria, a merging process is performed. The necessity of this process is that such segments are spatially short and color consistent with the main run. Merging them into the adjacent main run can restore the original continuous structure and eliminate the breakage caused by coding perturbation, thereby making the run length sequence features closer to the ideal state and improving the stability of subsequent matching.

[0055] When a fragment fails to meet the above criteria, it is removed. This is necessary because such fragments may contain non-target colors or abnormal information. Retaining or incorrectly merging them would introduce additional noise and interfere with the pattern matching process. By removing these fragments, noisy data can be filtered to some extent, thereby improving the overall data quality.

[0056] Performing safety-critical RLE pattern signature matching after rule aggregation is necessary because rule aggregation structurally corrects the original trip sequence, making the input data more stable and consistent in terms of trip length distribution, color continuity, and local structure. Matching based on this processing result can significantly reduce the probability of matching failure due to local interruptions or noise, thereby improving the accuracy and robustness of the initial identification.

[0057] Optionally, the step of performing a first verification on the above-mentioned first identification result based on vehicle operating status data to obtain a first verification conclusion includes:

[0058] Send diagnostic requests to the external control unit corresponding to the safety-critical information;

[0059] After the diagnostic request is sent, the response from the external control unit is captured and the response status corresponding to the diagnostic request is parsed out; the response status includes explicit fault, explicit no fault, system busy, stale data, or no response;

[0060] Continuously collect vehicle operating status data, including vehicle speed, engine load, and ambient temperature, and maintain a real-time updated vehicle status parameter table;

[0061] Maintain event logs for a preset time period; the event logs store the first identification results, the response status of external control units and fluctuations of key operating parameters in the most recent period, and identify recurring, continuous or aggravated abnormal patterns;

[0062] When the response status is system busy, outdated data, or no response, the current safety risk probability score is calculated based on the first identification result, response status, vehicle operation status data, and event log, and a preset adaptive warning display strategy is adopted according to the safety risk probability score as the first verification conclusion.

[0063] When the response status is a clear fault, the highest priority confirmed safety warning is displayed as the first verification conclusion;

[0064] When the response status is clearly fault-free, the first identification result is marked as a low-risk false alarm candidate and downgraded to display or not display. At the same time, the corresponding event is recorded in the event log as the basis for subsequent statistical calibration and as the first verification conclusion.

[0065] Specifically, diagnostic requests are sent to external control units corresponding to safety-critical information to proactively acquire system status or fault information related to that information. External control units can be understood as electronic control units within the vehicle responsible for specific functions (such as engine control units, brake control units, etc.). After a diagnostic request is sent, the system captures the response from the external control unit and parses the response status corresponding to the diagnostic request. The response status is the feedback from the external control unit to the diagnostic request, and can include various types such as "definite fault," "definite no fault," "system busy," "outdated data," or "no response," aiming to characterize the current operating status or diagnostic result of the external control unit. In practical applications, the system continuously collects vehicle operating status data, such as vehicle speed, engine load, and ambient temperature, and maintains a real-time updated vehicle status parameter table. This data provides real-time information on the vehicle's current operating environment and conditions, serving as an important basis for verification and judgment. Simultaneously, the system also maintains an event log for a preset time period. Event logs are used to store the first identification results, the response status of external control units, and the fluctuations of key operating parameters in a recent period. Their purpose is to identify whether there are recurring, continuous, or aggravated abnormal patterns through historical data analysis, thereby providing contextual information in the time dimension for current verification.

[0066] When the parsed response status is "System Busy," "Outdated Data," or "No Response," since the external control unit cannot provide a clear diagnostic result, the system will comprehensively calculate the current safety risk probability score based on the first identification result, the current response status, continuously collected vehicle operating status data, and event logs. This score is a quantitative indicator used to assess the probability that the safety-critical information is a real risk under the current circumstances. Based on the calculated safety risk probability score, the system will adopt a preset adaptive warning display strategy, such as adjusting the warning level, color, or display method, as the first verification conclusion. When the response status is "Clear Fault," it indicates that the external control unit has confirmed the existence of a fault. At this time, the system will directly display the highest priority confirmed safety warning as the first verification conclusion to ensure that the driver can immediately notice and take appropriate measures. When the response status is "Clear No Fault," it indicates that the external control unit has confirmed no fault. At this time, the system will mark the first identification result as a low-risk false alarm candidate and perform downgraded display or no display processing, while recording the corresponding event in the event log as a basis for subsequent statistical calibration, as the first verification conclusion. This is intended to avoid unnecessary warnings and provide data support for subsequent system optimization and false alarm rate calibration.

[0067] Optionally, a second identification is performed on the local region containing security-critical information in the data stream. The second identification is based on the relative color relationship and structural characteristics between pixels within the local region, and the steps to obtain the second identification conclusion include:

[0068] Background separation is performed on a local area containing safety-critical information. The background separation process includes: identifying a preset color area within the local area that is related to the safety-critical information; distinguishing the preset color area from the surrounding background area according to the boundary of the preset color area; the rule-based distinction is to segment the target area and the background area according to the preset color area boundary based on the preset rule.

[0069] Based on the results of rule-based differentiation, the relative brightness difference and relative hue difference between pixels within the safety-critical information region are calculated, and the travel length distribution within the safety-critical information region is analyzed.

[0070] Based on the relative brightness difference, relative hue difference, and travel length distribution among pixels within the safety-critical information area, calculate the differences in color contrast, boundary sharpness, and structural complexity between the safety-critical information area and the background area.

[0071] The differences in color contrast, boundary clarity, and structural complexity are matched with a preset safety-critical information template to obtain a second identification conclusion.

[0072] Specifically, background separation of local areas containing safety-critical information refers to the refined processing of the local image area containing the information after initial identification of potential safety-critical information. Identifying the preset color regions related to safety-critical information within the local area can be understood as quickly locating the possible color range of the safety-critical information, such as red warnings or yellow alerts, using color thresholds, color clustering, or preset color lookup tables. Distinguishing the preset color regions from the surrounding background based on their boundaries involves using image processing algorithms (such as edge detection, morphological operations, or connected component analysis) to determine the precise boundaries of the potential critical color regions after identification, and logically segmenting the target area (i.e., the safety-critical information region) from the surrounding background region based on these boundaries, thereby avoiding interference from background noise.

[0073] Furthermore, based on the results of rule-based differentiation, the relative brightness difference and relative hue difference between pixels within the safety-critical information region are calculated, and the run-length distribution within the safety-critical information region is analyzed. The relative brightness difference and relative hue difference are used to quantify the uniformity or gradation characteristics of colors within the region. For example, the brightness and hue differences within a solid-color icon will be very small, while those within a textured or gradient icon will be much larger. The run-length distribution describes the length characteristics of consecutive segments of the same color pixels in the RLE-encoded image, which is crucial for identifying icons with specific geometric shapes or line features.

[0074] Based on this, the differences in color contrast, boundary sharpness, and structural complexity between the safety-critical information area and the background area are calculated according to the relative brightness difference, relative hue difference, and travel length distribution among pixels within the safety-critical information area. Color contrast measures the visual distinction between the target area and the background; boundary sharpness assesses the sharpness of the target area's edges, helping to eliminate blurry or incomplete images; and structural complexity reflects the complexity of the texture and shape within the target area. For example, a simple warning symbol and a complex background pattern will have significantly different structural complexity.

[0075] Finally, the differences in color contrast, boundary sharpness, and structural complexity are matched with preset safety-critical information templates to obtain a second identification conclusion. Safety-critical information templates are pre-stored sets of features used to characterize various safety-critical information. For example, a template for a low fuel warning icon might include its typical red color, circular border, and structural complexity of specific internal symbols. By comparing the calculated features with these templates, it can be determined whether a local area indeed contains specific safety-critical information.

[0076] Optionally, the steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include:

[0077] Obtain the vehicle's current driving conditions; driving conditions include vehicle speed, vehicle acceleration, braking status, and steering angle;

[0078] The weights of the first identification result, the response status of the external control unit, vehicle operating status data, and event log in the scoring calculation are adjusted according to the driving conditions.

[0079] Adjust risk thresholds according to driving conditions;

[0080] Based on the adjusted weights and risk thresholds, and using the first identification result, response status, vehicle operating status data, and event logs, the current safety risk probability score is calculated.

[0081] Specifically, driving conditions refer to the vehicle's operating state at a specific point in time or within a specific time period, which can be comprehensively characterized by parameters such as vehicle speed, vehicle acceleration, braking status, and steering angle. Vehicle speed can be understood as the instantaneous speed of the vehicle; vehicle acceleration reflects the drastic change in vehicle speed; braking status indicates whether the vehicle is in the process of braking and its braking intensity; and steering angle represents the vehicle's steering intention and steering amplitude. These parameters collectively depict the vehicle's dynamic behavior pattern. Furthermore, adjusting the weights of various information sources in the scoring calculation based on driving conditions means dynamically changing the proportion of the first identification result, the response status of external control units, vehicle operating status data, and event logs in calculating the safety risk probability score based on the vehicle's current driving environment and operational behavior. For example, under high-speed driving conditions, operating status data such as vehicle speed and acceleration may be given higher weight to emphasize their importance in risk assessment; while under low-speed or stationary conditions, the weight of historical abnormal patterns in the event log may be relatively increased. Furthermore, adjusting the risk threshold based on driving conditions refers to dynamically setting or modifying the critical value used to determine the risk level based on the vehicle's real-time operating status. For example, in high-risk driving conditions such as emergency braking or high-speed cornering, the risk threshold can be appropriately lowered, making the system more sensitive to potential risks and triggering warnings earlier; while in smooth driving or parking conditions, the risk threshold can be appropriately raised to avoid false alarms caused by oversensitivity.

[0082] Optionally, the steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include:

[0083] Read the first identification result, the response status of the external control unit, vehicle operating status data, and event log;

[0084] The dynamics of driving conditions are assessed based on the vehicle speed change rate, steering angle change rate, and braking intensity. The dynamics are a quantitative indicator that characterizes the degree of drastic change in driving conditions within a preset time window. The dynamics are determined by at least one or more of the speed change rate, acceleration fluctuation amplitude, or steering angle fluctuation amplitude, and are used to trigger the linkage adjustment of timeliness weight, accuracy weight, and stability threshold.

[0085] Based on the degree of dynamism, timeliness weights and accuracy weights are assigned to the first identification result, response status, vehicle operation status data and event logs;

[0086] The first identification result, response status, vehicle operation status data and event log are used as various information sources;

[0087] For each information source, timestamp calibration and data integrity check are performed to obtain timestamp calibration results and data integrity check results;

[0088] For information sources whose timeliness is below a preset threshold due to timestamp calibration results or whose data integrity is insufficient due to data integrity checks, the contribution of the corresponding information source in the scoring calculation is reduced to calibrate the information source and obtain calibrated information sources. The contribution value characterizes the degree of participation of the corresponding information source's structured data packet in data fusion and consistency verification, as well as in the calculation of security risk probability scores. The contribution value is reduced when timestamp alignment deviation exceeds the allowable range or when the data integrity check is abnormal. The contribution value is jointly determined by the timeliness weight and the accuracy weight.

[0089] Based on the timeliness weight, accuracy weight, and various calibrated information sources, the current security risk probability score is calculated.

[0090] Specifically, before calculating the safety risk probability score, the system first needs to acquire or receive the initial identification result, the response status of the external control unit, vehicle operating status data, and event logs from within the system. These data constitute the basic information sources for assessing safety risks. The dynamic level of the driving condition is assessed by analyzing parameters such as the rate of change of vehicle speed, the rate of change of steering angle, and braking intensity. For example, when the vehicle speed changes drastically in a short period, the steering angle changes rapidly, or the braking intensity is high, the driving condition can be considered to be in a highly dynamic state. The dynamic level is a quantitative indicator that can be determined by one or more combinations of the above parameters. Its purpose is to reflect the complexity and rate of change of the driving environment, and based on this, dynamically adjust the weights and thresholds of various information items in subsequent scoring calculations. Furthermore, based on the assessed dynamic level of the driving condition, the system assigns different timeliness and accuracy weights to different information sources, namely the initial identification result, response status, vehicle operating status data, and event logs. For example, in highly dynamic conditions such as high speeds or sharp turns, data with higher real-time requirements (such as vehicle operating status data) may be assigned a higher timeliness weight, while the weight of historical data (such as event logs) may be correspondingly reduced. Furthermore, to ensure the quality of information sources participating in the scoring calculation, each information source undergoes timestamp calibration and data integrity checks. Timestamp calibration aims to ensure the time synchronization of all information sources, avoiding misjudgments due to data delays or out-of-order delivery. Data integrity checks are used to verify the reliability and integrity of the data, such as checking for corrupted or missing data packets. When the timestamp calibration result shows that the timeliness of an information source is below a preset threshold, or the data integrity check result shows insufficient data integrity, the contribution of that information source to the safety risk probability score calculation will be reduced. The contribution reflects the degree of participation of the information source in data fusion, consistency verification, and the final score calculation. In this way, information sources can be effectively calibrated, reducing the negative impact of low-quality or outdated data on the accuracy of the scoring. Finally, combining the dynamically adjusted timeliness weight, accuracy weight, and the calibrated information sources, the system will calculate the current safety risk probability score.

[0091] Optionally, the steps of assigning timeliness and accuracy weights to the first identification result, response status, vehicle operating status data, and event logs based on the degree of dynamism include:

[0092] Obtain the vehicle speed change rate, steering angle change rate, and braking intensity;

[0093] Calculate the trend strength index of speed change rate, steering angle change rate and braking intensity;

[0094] When the trend strength index exceeds the preset dynamic switching threshold, the pre-switching of the driving condition dynamic level is triggered; the pre-switching includes increasing the timeliness and accuracy weights associated with high dynamic conditions and decreasing the timeliness and accuracy weights associated with smooth driving conditions.

[0095] When the trend strength index remains below the dynamic decline threshold, the timeliness and accuracy weights associated with high dynamic conditions are reduced, while the timeliness and accuracy weights associated with smooth driving conditions are restored.

[0096] Specifically, acquiring vehicle speed change rate, steering angle change rate, and braking intensity refers to the system's real-time monitoring of vehicle operating data, including the instantaneous rate of change of vehicle speed, the rate of change of steering wheel rotation angle, and the change in brake pedal pressure or braking force. These parameters are key indicators characterizing the vehicle's current motion state and the driver's operational intentions. For example, the speed change rate can be obtained through differential calculation of vehicle speed sensor data; the steering angle change rate can be calculated from steering angle sensor data; and braking intensity can be provided by brake pedal position sensor or brake pressure sensor.

[0097] The trend intensity index, which calculates the rate of change of speed, the rate of change of steering angle, and the braking intensity, can be understood as a comprehensive analysis of the vehicle dynamic parameters obtained above, to quantify the severity or trend of the current driving condition. The trend intensity index can be a weighted average or a fusion calculation of these parameters using a machine learning model. Its purpose is to provide a single quantitative value that reflects the overall dynamics of the driving condition. For example, when the rate of change of vehicle speed, the rate of change of steering angle, and the braking intensity are all at high levels, the trend intensity index will indicate that the current driving condition has high dynamism.

[0098] In practical applications, when the trend intensity index exceeds a preset dynamic switching threshold, the system will trigger a pre-emptive switch in the driving condition's dynamic level. The dynamic switching threshold is a pre-set value used to distinguish between a relatively stable driving condition and a highly dynamic one. Pre-emptive switching specifically refers to the system proactively adjusting the weighting parameters used for calculating the safety risk probability score. Specifically, the timeliness and accuracy weights associated with highly dynamic conditions will be increased. This means the system will place greater emphasis on the latest data and data that may sacrifice some accuracy but can be obtained quickly, ensuring a rapid response in emergencies. Simultaneously, the timeliness and accuracy weights associated with stable driving conditions will be decreased to avoid over-reliance on data that may be outdated or unsuitable for the current highly dynamic environment during dynamic changes.

[0099] Furthermore, when the trend strength indicator remains below the dynamic fallback threshold, the system will reduce the timeliness and accuracy weights associated with high-dynamic conditions and restore the timeliness and accuracy weights associated with stable driving conditions. The dynamic fallback threshold is typically lower than the dynamic switching threshold to provide a certain degree of lag or "hysteresis effect," preventing unnecessary weight fluctuations when the system frequently switches between dynamic and stable driving conditions. This aims to ensure that the system can smoothly adjust the weights back to a configuration that prioritizes data accuracy and stability when the driving condition recovers from high-dynamic to stable, thereby maintaining the long-term reliability of risk assessment while ensuring responsiveness.

[0100] Optionally, the steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include:

[0101] Get the vehicle's current driving mode;

[0102] The weights of the first identification result, the response status of the external control unit, the vehicle operating status data, and the event log in the scoring calculation are adjusted according to the driving mode.

[0103] Based on the first identification result, response status, vehicle operation status data, and event logs, and according to the adjusted weights, the current safety risk probability score is calculated.

[0104] Specifically, obtaining the vehicle's current driving mode refers to the system acquiring information about the vehicle's current driving mode in real time. A driving mode can be understood as a pre-set operating configuration by the vehicle manufacturer that affects vehicle dynamics and the behavior of assistance systems, such as "Economy Mode," "Comfort Mode," "Sport Mode," or "Snow Mode." These modes are typically broadcast by the vehicle control unit via an onboard bus (such as the CAN bus), and the system can obtain them by reading or querying the corresponding messages. The purpose is to provide important contextual information for subsequent risk assessments.

[0105] The weighting of the first identification result, external control unit response status, vehicle operating status data, and event logs in the scoring calculation, based on the driving mode, refers to dynamically adjusting the relative importance of each information source (including the first identification result, external control unit response status, vehicle operating status data, and event logs) used to calculate the safety risk probability score according to the currently acquired driving mode. For example, in "Sport mode," the system may assign higher weight to certain performance-related warnings (such as engine overheating) and lower weight to some minor warnings that do not affect driving performance; while in "Economy mode," it may assign higher weight to warnings related to fuel efficiency or environmental protection. The weighting adjustment can be achieved through a preset lookup table, rule-based logic, or machine learning model to ensure that the risk assessment matches the characteristics of the driving mode.

[0106] In practical applications, calculating the current safety risk probability score based on the initial identification result, response status, vehicle operating status data, and event logs, and according to adjusted weights, involves acquiring and adjusting the weights of various information sources, inputting these weighted information sources into a pre-defined risk assessment model, and comprehensively calculating the current safety risk probability score. This score is a quantitative indicator used to characterize the probability or severity of a safety event occurring under the current vehicle operating state. By introducing driving modes for weight adjustment, the final risk score can be made more closely aligned with actual driving scenarios and driver expectations.

[0107] Optionally, the step of calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs includes:

[0108] Obtain the current driving status of the vehicle;

[0109] Determine whether the driving status indicates that the vehicle is at a speed below the preset speed or is stopped;

[0110] When the vehicle is below the preset speed or stopped, a detection request is sent to the external control unit to obtain the response status corresponding to the detection request, and the response status corresponding to the diagnostic request is replaced with the response status corresponding to the detection request.

[0111] Continuously read the regular operating status messages from the external control unit on the vehicle communication bus, and analyze the data update frequency and value fluctuations of the regular operating status messages;

[0112] Based on the data update frequency and numerical fluctuations, identify whether there are changes in the operating status data of other vehicles;

[0113] When the response status indicates that the external control unit continues to return an uncertain response and other vehicle operating status data remain unchanged, the current status is marked as expected uncertainty under low dynamic conditions.

[0114] When anticipated uncertainties occur under low dynamic operating conditions, reduce the negative impact of uncertain responses returned by external control units on the safety risk probability score, and increase the weight of other vehicle operating status data in the score;

[0115] The stability threshold is increased based on the duration of the uncertain response. The stability threshold is the tolerance duration threshold used to determine the uncertain response of the external control unit under low dynamic conditions. Increasing the stability threshold is used to delay the triggering of negative risk accumulation, so as to avoid the expected uncertainty caused by stopping or low speed being misjudged as high risk.

[0116] Based on the first identification result, response status, vehicle operation status data and event logs, and according to the adjusted weights and stability thresholds, the current safety risk probability score is calculated.

[0117] Specifically, obtaining the vehicle's current driving status refers to acquiring information such as the vehicle's speed, acceleration, and gear position in real time through onboard sensors or vehicle bus data to comprehensively understand the vehicle's current movement. Determining whether the driving status indicates the vehicle is below a preset speed or stationary can be done by comparing the acquired vehicle speed with a preset speed threshold. For example, when the vehicle speed is below 5 km / h or zero, it can be determined to be in a low-speed or stationary state.

[0118] When the vehicle is below a preset speed or stationary, the system sends a probe request to an external control unit. This probe request is a lightweight query designed to confirm the basic operational status of the external control unit, rather than delving into specific fault information like a diagnostic request. By obtaining the response status of the probe request and replacing it with the response status of the diagnostic request, misjudgments caused by delayed responses or the return of general, uncertain information from the external control unit in low-dynamic conditions can be avoided.

[0119] Meanwhile, the system continuously reads routine operating status messages from external control units on the vehicle communication bus, such as engine speed, door status, and light status. By analyzing the data update frequency and value fluctuations of these messages, it can identify whether there are changes in other vehicle operating status data. For example, if the engine speed is stable, the doors are closed, the lights are working properly, and these data update frequencies are normal, it indicates that other vehicle systems are operating smoothly.

[0120] When the response status indicates that the external control unit continues to return an uncertain response (such as "system busy" or "no response") and other vehicle operating status data remain unchanged, the system will mark the current state as expected uncertainty under low dynamic conditions. This means that the system recognizes the current uncertain response as a normal phenomenon due to the vehicle being in low dynamic conditions, rather than an actual fault indication.

[0121] When anticipated uncertainties arise under low-dynamic operating conditions, the system reduces the negative impact of uncertain responses returned by external control units on the safety risk probability score and increases the weight of other vehicle operating status data in the score. For example, the risk weight of uncertain responses can be reduced from 0.8 to 0.2, while the weight of other stable operating status data can be increased from 0.5 to 0.8 to more accurately reflect the actual risk.

[0122] Furthermore, the system increases the stability threshold based on the duration of the uncertain response. The stability threshold is a tolerance threshold used to determine the duration of uncertain responses from external control units under low-dynamic conditions. Increasing the stability threshold means the system extends the waiting time, delaying the triggering of negative risk accumulation, thereby preventing anticipated uncertainties caused by stopping or low speeds from being misjudged as high risk. For example, if the normal stability threshold is 5 seconds, it can be increased to 15 seconds under low-dynamic conditions.

[0123] This application also discloses a color screen instrument image classification system based on RLE decompression, used to perform color screen instrument image classification based on RLE decompression, combined with... Figure 3 As shown, the color screen instrument image classification system 1 based on RLE decompression includes:

[0124] The preliminary identification execution module 11 is used to perform preliminary identification of the data stream based on the preset safety key RLE mode signature during the decompression process of the RLE compressed image data stream. The preset safety key RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety key display area. The signature template includes at least a key area identifier, a key color category identifier, a run length sequence feature, and a matching threshold configuration.

[0125] The identification and verification conclusion module 12 is used to output a first identification result and start a preset high-priority verification process when safety-critical information is identified during the preliminary identification process. The high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety-critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on the local area containing safety-critical information in the data stream, the second identification being based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion.

[0126] The identification result processing module 13 is used to make a final confirmation based on the first identification result, the first verification conclusion and the second identification conclusion. If the safety key information category indicated by the first identification result is consistent with the safety key information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the safety key information category indicated by the first identification result is inconsistent with the safety key information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

[0127] Specifically, in the above system structure, the setup of each module and their collaborative relationship have clear technical necessity, and together they constitute a low-latency and high-reliability identification mechanism suitable for RLE data stream scenarios.

[0128] The initial recognition module is configured to perform pattern matching synchronously during the decompression of the RLE compressed image data stream. This configuration is necessary because RLE data streams are based on runs as the basic unit. If full decoding is performed before recognition, it introduces additional storage and latency overhead. Furthermore, errors can propagate during the decoding stage if data anomalies exist. By directly matching the run sequences during decompression, key features can be extracted while the data is still in its compressed representation stage, reducing redundant processing caused by intermediate data expansion. In addition, using hardware logic to implement scanning and matching operations allows run-level comparisons to be completed within a clock cycle, avoiding the latency accumulation caused by pixel-by-pixel processing in software. The firmware and hardware collaboration decouples signature template updates from the execution of matching logic, thus maintaining system configurability while ensuring processing efficiency.

[0129] The identification and verification conclusion module initiates a high-priority verification process after the initial identification results are generated. This is necessary because the initial identification based on RLE travel features is a rapid screening mechanism that focuses on structural matching rather than semantic confirmation, and there is still a risk of misjudgment when compression perturbations or data anomalies exist. Therefore, an independent verification path is needed to constrain the identification results. By obtaining vehicle operating status data from an external control unit and performing the first verification, a correspondence between the image recognition results and the actual operating status can be established, so that the identification process no longer relies solely on the image data itself, thus avoiding inconsistencies between displayed information and the true state. Simultaneously, by performing a second identification on local areas, using the relative relationships and structural characteristics between pixels for judgment, the instability of RLE encoding in color representation and travel segmentation can be compensated for, enabling the system to retain structural-level recognition capabilities even when local data is distorted. Organizing communication processing, data verification, and local analysis in parallel within the same module can shorten the response time of the verification path and ensure processing continuity in high-priority scenarios.

[0130] The recognition result processing module performs final confirmation based on multi-source verification results. This is necessary because conclusions from travel matching, operational status verification, and local structural analysis differ in their judgment criteria. Directly using a single result for output would compromise both real-time performance and reliability. By constructing a unified decision logic and constraining the consistency of multi-source results, information from different dimensions can be fused, thereby improving the stability of the judgment. When all verification results are consistent, it indicates a closed relationship between image features, physical state, and structural information, resulting in a highly reliable output. When inconsistencies exist, a rollback operation prevents uncertain results from entering the display path, reducing the risk of false alarms interfering with the user. Furthermore, integrating risk assessment and display control into the same processing unit allows for direct association of the output strategy during the judgment process, avoiding delays caused by multi-level transmission.

[0131] From an overall structural perspective, this system divides the recognition process into three stages: "decompression and synchronous recognition, multi-source verification, and consistency decision-making." Each stage is responsible for feature extraction, authenticity verification, and result constraint, respectively. The necessity of this hierarchical structure lies in improving the reliability of the results by progressively increasing constraints while ensuring recognition speed, thus avoiding performance conflicts arising from simultaneously handling high speed and high accuracy requirements in a single processing stage. Compared to processing methods that classify solely based on complete images, this structure utilizes the encoding characteristics of RLE data streams to complete key feature localization before the data is unfolded. Furthermore, through a dual verification mechanism of external data and local structure, the recognition results are cross-constrained, thereby maintaining stable recognition and output of security-critical information even when data disturbances or inconsistencies exist.

[0132] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A color screen instrument image classification method based on RLE decompression, characterized in that, include: During the decompression process of the RLE compressed image data stream, the data stream is initially identified based on the preset safety-critical RLE mode signature; the preset safety-critical RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety-critical display area, and the signature template includes at least a key area identifier, a key color category identifier, a run length sequence feature, and a matching threshold configuration. When critical safety information is identified during the initial identification process, the first identification result is output and a preset high-priority verification process is initiated. The high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on a local area in the data stream containing the safety critical information, wherein the second identification is based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion. Based on the first identification result, the first verification conclusion, and the second identification conclusion, a final confirmation is made. If the category of safety critical information indicated by the first identification result is consistent with the category of safety critical information indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the category of safety critical information indicated by the first identification result is inconsistent with the category of safety critical information indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

2. The image classification method for color screen instruments based on RLE decompression according to claim 1, characterized in that, The step of initially identifying the data stream based on a preset security key RLE mode signature during the decompression process of the RLE compressed image data stream includes: The data stream of the RLE compressed image is subjected to rule aggregation. The rule aggregation process includes: at the start of rule aggregation, if a segment of the expected main color is interrupted by a segment within the window, rule aggregation judgment is initiated; wherein, the window is a preset decompression scan range; the segment is the length of consecutive pixels of the same color in the RLE; the segment is a segment of the segment that has been interrupted or split; after initiating rule aggregation judgment, it is determined whether the length of the interrupted segment is lower than the segment length threshold, and a segment length judgment result is obtained; it is determined whether the hue difference and saturation difference between the color of the interrupted segment and the expected main color in the color space are both less than the difference threshold, and a color difference judgment result is obtained; when both the segment length judgment result and the color difference judgment result are yes, the interrupted segment is merged or removed according to preset rules and reconstructed into a continuous segment; wherein, the process of merging or removing the interrupted segment according to preset rules is: when the length of the interrupted segment is lower than the preset length threshold and the color difference with the adjacent main segment is lower than the preset color threshold, the interrupted segment is merged into the adjacent main segment; otherwise, the interrupted segment is removed when generating a continuous segment. The continuous routes obtained after rule aggregation are matched with the preset safety-critical RLE pattern signature to perform preliminary identification of the data stream.

3. The image classification method for color screen instruments based on RLE decompression according to claim 1, characterized in that, The step of performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion includes: Send diagnostic requests to the external control unit corresponding to the safety-critical information; After the diagnostic request is sent, the response from the external control unit is captured and the response status corresponding to the diagnostic request is parsed out; the response status includes clear fault, clear no fault, system busy, stale data, or no response; Continuously collect vehicle operating status data; the vehicle operating status data includes vehicle speed, engine load and ambient temperature, and maintain a real-time updated vehicle status parameter table; Maintain event logs for a preset time period; the event logs store the first identification results, the response status of the external control unit and the fluctuation of key operating parameters in the most recent period, and identify recurring, continuous or aggravated abnormal patterns; When the response status is system busy, outdated data, or no response, the current safety risk probability score is calculated based on the first identification result, response status, vehicle operation status data, and event log, and a preset adaptive warning display strategy is adopted according to the safety risk probability score as the first verification conclusion. When the response status is a clear fault, the highest priority confirmed safety warning is displayed as the first verification conclusion; When the response status is clearly fault-free, the first identification result is marked as a low-risk false alarm candidate and downgraded display or non-display processing is performed. At the same time, the corresponding event is recorded in the event log as the basis for subsequent statistical calibration and as the first verification conclusion.

4. The image classification method for color screen instruments based on RLE decompression according to claim 1, characterized in that, The step of performing a second identification on a local region in the data stream containing the security-critical information, wherein the second identification is based on the relative color relationship and structural characteristics between pixels within the local region, and obtaining a second identification conclusion includes: Background separation is performed on a local area containing the safety-critical information. The background separation process includes: identifying a preset color area within the local area that is related to the safety-critical information; and distinguishing the preset color area from the surrounding background area according to the boundary of the preset color area. The rule-based distinction is to segment the target area and the background area according to the preset color area boundary based on a preset rule. Based on the results of rule-based differentiation, the relative brightness difference and relative hue difference between pixels within the safety-critical information region are calculated, and the travel length distribution within the safety-critical information region is analyzed. Based on the relative brightness difference, relative hue difference, and travel length distribution among pixels within the safety-critical information area, calculate the differences in color contrast, boundary sharpness, and structural complexity between the safety-critical information area and the background area. The differences in color contrast, boundary clarity, and structural complexity are matched with a preset safety-critical information template to obtain a second identification conclusion.

5. The image classification method for color screen instruments based on RLE decompression according to claim 3, characterized in that, The steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include: Obtain the current driving conditions of the vehicle; the driving conditions include vehicle speed, vehicle acceleration, braking status and steering angle; The weights of the first identification result, the response status of the external control unit, the vehicle operating status data, and the event log in the scoring calculation are adjusted according to the driving conditions. Adjust the risk threshold according to the driving conditions described; Based on the adjusted weights and risk thresholds, and using the first identification result, response status, vehicle operating status data, and event logs, the current safety risk probability score is calculated.

6. The image classification method for color screen instruments based on RLE decompression according to claim 3, characterized in that, The steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include: Read the first identification result, the response status of the external control unit, vehicle operating status data, and event log; The dynamic degree of driving conditions is evaluated based on the vehicle speed change rate, steering angle change rate, and braking intensity. The dynamic degree is a quantitative indicator that characterizes the degree of drastic change in driving conditions within a preset time window. The dynamic degree is determined by at least one or more of the speed change rate, acceleration fluctuation amplitude, or steering angle velocity fluctuation amplitude, and is used to trigger the linkage adjustment of timeliness weight, accuracy weight, and stability threshold. Based on the dynamic level, timeliness weights and accuracy weights are assigned to the first identification result, the response status, the vehicle operation status data, and the event log; The first identification result, the response status, the vehicle operating status data, and the event log are used as various different information sources; For each information source, timestamp calibration and data integrity check are performed to obtain timestamp calibration results and data integrity check results; For information sources whose timeliness is below a preset threshold due to timestamp calibration results or whose data integrity is insufficient due to data integrity checks, the contribution of the corresponding information source in the scoring calculation is reduced to calibrate the information sources and obtain calibrated information sources. The contribution degree characterizes the degree of participation of the corresponding information source's structured data packets in data fusion and consistency verification, as well as in the calculation of security risk probability scores. The contribution degree is reduced when the timestamp alignment deviation exceeds the allowable range or the data integrity check is abnormal. The contribution degree is jointly determined by a timeliness weight and an accuracy weight. Based on the timeliness weight, accuracy weight, and various calibrated information sources, the current security risk probability score is calculated.

7. The image classification method for color screen instruments based on RLE decompression according to claim 6, characterized in that, The step of assigning timeliness weights and accuracy weights to the first identification result, the response state, the vehicle operating status data, and the event log based on the dynamic degree includes: Obtain the vehicle speed change rate, steering angle change rate, and braking intensity; Calculate the trend strength index of the speed change rate, steering angle change rate, and braking intensity; When the trend intensity index exceeds the preset dynamic switching threshold, the pre-switching of the driving condition dynamic level is triggered; the pre-switching includes increasing the timeliness weight and accuracy weight associated with high dynamic conditions, and decreasing the timeliness weight and accuracy weight associated with stable driving conditions. When the trend strength index remains below the dynamic decline threshold, the timeliness and accuracy weights associated with high dynamic conditions are reduced, while the timeliness and accuracy weights associated with smooth driving conditions are restored.

8. The image classification method for color screen instruments based on RLE decompression according to claim 3, characterized in that, The steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include: Get the vehicle's current driving mode; The weights of the first identification result, the response status of the external control unit, the vehicle operating status data, and the event log in the scoring calculation are adjusted according to the driving mode. Based on the first identification result, response status, vehicle operation status data, and event logs, and according to the adjusted weights, the current safety risk probability score is calculated.

9. The image classification method for color screen instruments based on RLE decompression according to claim 3, characterized in that, The steps for calculating the current safety risk probability score based on the first identification result, response status, vehicle operating status data, and event logs include: Obtain the current driving status of the vehicle; Determine whether the driving status indicates that the vehicle is at a speed below the preset speed or is stopped; When the vehicle is below the preset speed or stopped, a detection request is sent to the external control unit to obtain the response status corresponding to the detection request, and the response status corresponding to the diagnostic request is replaced with the response status corresponding to the detection request. Continuously read the regular operating status messages from the external control unit on the vehicle communication bus, and analyze the data update frequency and value fluctuations of the regular operating status messages; Based on the data update frequency and numerical fluctuations, identify whether there are changes in the operating status data of other vehicles; When the response status indicates that the external control unit continues to return an uncertain response and other vehicle operating status data remain unchanged, the current status is marked as expected uncertainty under low dynamic conditions. When anticipated uncertainties occur under low dynamic operating conditions, reduce the negative impact of uncertain responses returned by external control units on the safety risk probability score, and increase the weight of other vehicle operating status data in the score; The stability threshold is increased based on the duration of the uncertain response. The stability threshold is a tolerance duration threshold used to determine the uncertain response of the external control unit under low dynamic operating conditions. Increasing the stability threshold is used to delay the triggering of negative risk accumulation, so as to avoid the expected uncertainty caused by stopping or low speed being misjudged as high risk. Based on the first identification result, response status, vehicle operation status data and event logs, and according to the adjusted weights and stability thresholds, the current safety risk probability score is calculated.

10. A color screen instrument image classification system based on RLE decompression, used to perform color screen instrument image classification based on RLE decompression, characterized in that, include: The preliminary identification execution module is used to perform preliminary identification of the data stream of the RLE compressed image based on the preset safety key RLE mode signature during the decompression process. The preset safety key RLE mode signature is a signature template used to characterize the RLE run sequence shape of the safety key display area. The signature template includes at least a key area identifier, a key color category identifier, a run length sequence feature, and a matching threshold configuration. The identification and verification conclusion module is used to output the first identification result and start the preset high-priority verification process when safety-critical information is identified during the initial identification process. The high-priority verification process includes: sending a diagnostic request to an external control unit to obtain vehicle operating status data corresponding to the safety critical information, and performing a first verification on the first identification result based on the vehicle operating status data to obtain a first verification conclusion; performing a second identification on a local area in the data stream containing the safety critical information, wherein the second identification is based on the relative color relationship and structural characteristics between pixels in the local area to obtain a second identification conclusion. The identification result processing module is used to make a final confirmation based on the first identification result, the first verification conclusion, and the second identification conclusion. If the safety critical information category indicated by the first identification result is consistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is displayed with the highest priority. If the safety critical information category indicated by the first identification result is inconsistent with the safety critical information category indicated by the first verification conclusion and the second identification conclusion, the first identification result is not adopted and a rollback operation is performed.

Images