Vehicle-mounted camera image interference stripe elimination method, system and device based on aging state adaptation

Abstract

The application discloses a vehicle-mounted camera image interference stripe elimination method, system and equipment based on an aging state self-adaption, wherein the method comprises the following steps: S1: acquiring real-time image data collected by a camera and converting the real-time image data into a frequency spectrum; S2: periodically collecting a plurality of frequency spectrum and calculating a current interference frequency; S3: calling all historical interference frequency data stored locally by the camera, comparing the current interference frequency with the historical interference frequency through a first threshold value and a second threshold value; and S4: adjusting the parameters of a filter by using the updated interference frequency data and adaptively filtering the real-time image data. According to the application, the filtering characteristics are dynamically adjusted according to environmental changes, the filter is always configured in an optimal working state, the subsequently generated filter parameters can more accurately suppress existing interference, the dynamic changes of the interference have a quick response capability, the robustness of the system is improved, and the stability of the image quality is ensured.

Description

Technical Field

[0001] This invention relates to the field of camera image processing, and in particular to a method, system, and device for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation. Background Technology

[0002] Cameras are currently essential image sensors in various vehicles and engineering equipment, used to observe the working conditions of the surroundings and interior of the equipment. However, in practical use, embedded video acquisition systems such as security monitoring, industrial vision, and vehicle imaging often encounter interference sources in the composite video signals output by analog cameras due to the complex characteristics of the entire device. These interference sources typically arise from various complex situations, such as: 1. Switching frequency interference from the camera's own power supply and various switching frequency interferences from the display device's internal power supply; 2. Interference introduced from the vehicle's power supply and ground wire; 3. Frequency interference from other normal equipment; 4. Interference from the real-time changes in the various states of different electrical components in the vehicle; 5. Interference gradually introduced as the equipment ages, etc. Because multiple interference components can easily be mixed into the signal, it can lead to unclear or distorted image display, seriously affecting safety.

[0003] Traditional interference suppression schemes typically employ back-end digital image processing: after the video acquisition chip completes analog-to-digital conversion, an embedded CPU or FPGA performs algorithms such as median filtering, Wiener filtering, and wavelet denoising on the pixel data of each frame. However, for high-definition video streams with high real-time requirements, the computational load of these algorithms is enormous. Taking common ARM Cortex-A series embedded processors as an example, their processing power is insufficient to complete the complex filtering operations of millions of pixels per frame without losing frames, while also significantly increasing system cost, power consumption, and development difficulty.

[0004] Another existing technology typically involves setting up filters with fixed parameters in the image processing chain, that is, pre-setting one or more interference frequency points to continuously filter the acquired image. However, this fixed-parameter approach has obvious drawbacks: on the one hand, interference frequencies vary under different usage scenarios, and preset parameters are difficult to cover all situations; on the other hand, since interference sources may drift or be added as the device's usage status, environment, and time change, fixed filters often can only solve the current specific situation and cannot be dynamically adjusted to cope with different situations. Summary of the Invention

[0005] Therefore, to solve the above problems, the present invention provides a method, system and device for eliminating interference stripes in vehicle camera images based on aging state adaptation.

[0006] This invention is achieved through the following technical solution: An aging-adaptive method for eliminating interference stripes in vehicle-mounted camera images includes the following steps: S1: Acquire real-time image data captured by the camera and convert the real-time image data into a spectrogram; S2: Collect several spectrum graphs in a cycle, extract the highest peak values ​​in all spectrum graphs in each cycle, calculate the current interference frequency, and determine whether the current interference frequency is effective. S3: Retrieve all historical interference frequency data stored locally by the camera, compare the current interference frequency with the historical interference frequency using a first threshold and a second threshold. The first threshold is used to determine whether the current interference frequency is within the effective filtering range of the current filter, and the second threshold is used to determine the type of the current interference frequency, distinguishing whether the current interference frequency is a regular anomaly caused by device aging or other abnormal interference frequencies; and dynamically store or change the interference frequency data based on the current interference frequency type to achieve self-learning. S4: Adjust the filter parameters using the updated interference frequency data and adaptively filter the real-time image data.

[0007] Preferably, step S1 includes the following steps: S11: Acquire real-time image data; S12: Perform grayscale processing on each frame of the image; S13: Move the window to reset the detection image area; S14: Perform Fourier transform on the image of the detection region selected by the moving window in each frame to obtain the spectrogram.

[0008] Preferably, in step S13, when setting the detection image region for each frame, the control window moves to a random position and randomly generates the detection image region for the current image.

[0009] Preferably, step S2 includes the following steps: S21: Store the spectrograms into the cache queue in real time, and determine whether the number of spectrograms in the cache queue has reached the target number. If the target number has been reached, stop storing images and proceed to step S22. S22: Extract the top three highest peaks from all spectrograms in the cache queue; S23: Calculate the current interference frequency based on the three peak values; S24: Clear the spectrogram in the cache queue; S25: Determine whether the fluctuation value between the current interference frequency and the previously stored historical interference frequency exceeds the interference frequency threshold. If it is greater than the interference frequency threshold, the current interference frequency is determined to be valid, and proceed to step S3; if it is less than or equal to the interference frequency threshold, the current interference frequency is determined to be invalid, and no adjustment of filter parameters is required.

[0010] Preferably, step S23 includes the following steps: Calculate the percentage difference between the three peaks. If the percentage difference between a peak and the other two peaks exceeds 5%, then that peak is taken as an independent interference frequency. If the percentage difference between two or three peaks does not exceed 5%, then the average of these peaks is calculated as the interference frequency.

[0011] Preferably, in step S24, after clearing the spectrogram in the cache queue, a new spectrogram is stored for calculation in the next cycle.

[0012] Preferably, step S3 includes the following steps: S31: Retrieve the interference frequency data stored locally by the camera to obtain all historical interference frequencies; S32: Calculate the difference between the current interference frequency and each historical interference frequency, and determine whether the difference between the current interference frequency and each historical interference frequency is greater than the first threshold; if yes, proceed to step S33; if no, there is no need to update the interference frequency data. S33: Extract the historical interference frequency with the smallest difference from the current interference frequency among all historical interference frequencies, and calculate the difference between the current interference frequency and the historical interference frequency. Determine whether the difference is greater than the second threshold. If yes, store the current interference frequency as a new interference frequency in the interference frequency data. If no, overwrite the historical interference frequency with the one with the smallest difference with the current interference frequency.

[0013] An aging-adaptive image interference stripe elimination system for automotive cameras includes: The acquisition module includes at least one camera for acquiring real-time image data. The image processing module is used to convert real-time image data captured by each camera into a spectrogram; The local storage module is used to store all historical interference frequency data; The adaptive learning module performs local self-learning based on the current interference frequency and the historical interference frequencies in the local storage module, and dynamically stores and updates the interference frequency data in the local storage module based on the self-learning results. Filtering module, including filters; The adjustment module is used to configure the filter channel parameters of each channel based on the updated interference frequency data.

[0014] Preferably, it also includes a caching module, which includes a cache queue for caching the spectrograms after Fourier transform, records the number of spectrograms cached in the cache queue in real time, determines whether the number of spectrograms has reached the target number, and if the target number has been reached, outputs all spectrograms in the cache queue to the adaptive learning module, then clears the current spectrogram data, and continues to cache the spectrograms of the next period.

[0015] The device includes: at least one camera, an ADC chip, and an embedded processor; The camera is used to collect real-time image data; The ADC chip includes a digital filter, which is used to achieve precise separation of luminance and chrominance in composite video signals and suppression of out-of-band noise by setting specific coefficients under a 27MHz sampling clock, and transmits the filtered digital image data to the embedded processor in real time via the CSI bus. The embedded processor receives digital image frames from the ADC chip via the CSI bus, performs grayscale conversion, moving window selection, and spectrum analysis on the image, identifies the interference frequency that needs to be suppressed, and performs self-learning based on historical interference data to dynamically update the interference frequency database. Based on the learning results, it calculates the current filter coefficients, writes the configuration command back to the ADC chip via the I²C bus, and adjusts its internal filter parameters in real time.

[0016] The beneficial effects of the technical solution of this invention are mainly reflected in: 1. This invention uses a self-learning mechanism that combines real-time recorded interference frequencies with historical interference frequencies stored locally in the camera. It compares and analyzes the current interference frequency with all historical interference frequencies, and dynamically stores and modifies the interference frequency data based on the self-learning results. This adaptively adjusts the filter parameters, giving the system learning and memory capabilities. It can simultaneously adapt to slow drifts and periodic changes in interference frequencies, maintaining stable image quality even in complex electromagnetic environments. Each camera can independently adapt to its own interference characteristics even in aging conditions, making it suitable for scenarios with high image quality requirements and complex, ever-changing interference environments. Furthermore, by dynamically adjusting the filtering characteristics according to environmental changes, the filter is always configured in its optimal working state, allowing subsequently generated filter parameters to more accurately suppress existing interference. It exhibits rapid response to dynamic changes in interference, improving system robustness and ensuring image quality stability.

[0017] 2. This invention acquires images in real time and converts them into spectrograms. It then dynamically detects interference frequency components in the current image by identifying high-amplitude points in the spectrogram and determines in real time whether the current interference frequency needs processing. This suppresses only the effective interference with strong energy, avoiding the loss of image details caused by over-filtering. Thus, while effectively eliminating stripe interference, it preserves the original details and clarity of the image. Furthermore, by jointly analyzing abnormal high-amplitude points in the spectrograms of multiple frames, it can effectively suppress random noise interference in a single frame spectrum, improving the accuracy and stability of interference frequency detection.

[0018] 3. In the device of this invention, interference suppression operations are concentrated at the very front end of the signal link. All filtering operations are completed by dedicated hardware circuits inside the ADC chip. The embedded processor does not need to perform any pixel-level filtering algorithms on each frame of the image. It only needs to extract a small amount of image data for spectrum analysis to update the filter parameters. The computational load of this analysis is extremely small and it is performed in parallel with the real-time video acquisition, which greatly improves work efficiency. At the same time, the spectrum analysis results from the back end are fed back to the front-end hardware filter (the internal filter of the ADC chip) in real time for filtering from the input source. Therefore, interference data will not be mixed into the back end of the image, avoiding subsequent nonlinear distortion, spectrum aliasing and other problems, forming a closed-loop adaptive interference removal system. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to the present invention; Figure 2 This is a flowchart of a single calculation of interference frequency in the aging-adaptive vehicle camera image interference stripe elimination method of the present invention. Figure 3 This is a flowchart of adjusting filter parameters in the aging-adaptive image interference stripe elimination method for vehicle cameras in this invention; Figure 4 This is a schematic diagram showing the connection relationship between the camera, ADC chip, and embedded processor in the device of the present invention. Detailed Implementation

[0020] To make the objectives, advantages, and features of the present invention clearer and more detailed, the following non-limiting description of preferred embodiments will be illustrated and explained. These embodiments are merely typical examples of applying the technical solutions of the present invention; all technical solutions formed by equivalent substitutions or equivalent transformations fall within the scope of protection claimed by the present invention.

[0021] It should also be noted that in the description of the solution, the terms "center", "upper", "lower", "left", "right", "front", "rear", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of description and simplification, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0022] Furthermore, the terms "first" and "second" in this invention are used for descriptive purposes only and should not be construed as indicating or implying a ranking of importance, or implicitly specifying the number of technical features shown. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] An aging-adaptive method for eliminating interference stripes in vehicle camera images, such as... Figures 1-3 As shown, it includes the following steps: S1: Acquire real-time image data captured by the camera and convert the real-time image data into a spectrogram.

[0024] In some embodiments, such as Figure 2 As shown, step S1 includes the following steps: S11: Acquire real-time image data, wherein the real-time image data is an analog video signal acquired by the camera in real time; S12: Perform grayscale processing on each frame of the image; S13: Move the window to reset the detection image area; S14: Perform Fourier transform on the image of the detection region selected by the moving window in each frame to obtain the spectrogram; In one embodiment, steps S12-S14 are completed by an embedded processor. Before step S12, the analog video signal acquired in real time by the camera is converted into a digital signal and input into the embedded processor. Then, grayscale processing is performed on each frame of the image and the detection image region is selected. Finally, the spectrum is obtained by FFT calculation. The Fast Fourier Transform (FFT) algorithm is an existing calculation method and will not be described in detail here. In step S13, since the interference signal exhibits random rolling characteristics in the image, when setting the detection image region for each frame of the image, the control window is moved to a random position and the detection image region of the current image is randomly generated to adapt to the random rolling characteristics of the interference signal. At the same time, the sampling size can be adjusted according to the specific embedded processor (CPU) used, without exceeding its certain performance, as required. This will not be described in detail here.

[0025] S2: Collect several spectrum graphs in a cycle, extract the highest peak values ​​in all spectrum graphs in each cycle, calculate the current interference frequency, and determine whether the current interference frequency is effective.

[0026] like Figure 2 As shown, in some embodiments, step S2 includes the following steps: S21: Store the spectrograms in the buffer queue in real time, and determine whether the number of spectrograms in the buffer queue has reached the target number. If the target number has been reached, stop storing images and proceed to step S22; Figure 2 As shown, in a preferred embodiment, the target number of spectrograms is 100. When the number of spectrograms in the cache queue reaches 100, the input of spectrograms is temporarily stopped, and the 100 spectrograms in the cache queue are analyzed first to obtain the current interference frequency. S22: Extract the top three highest peaks from all spectrograms in the cache queue; wherein, in the preferred embodiment above, three highest peaks are found in 100 spectrograms, and these three peaks represent the abnormal signals in the 100 spectrograms; in other embodiments, the number of peaks extracted from the spectrograms can be adjusted according to actual needs, which will not be elaborated here. S23: Calculate the current interference frequency based on the three peak values; in a preferred embodiment, step S23 includes the following steps: Calculate the percentage difference between the three peaks. Specifically, compare each of the three peaks pairwise, and divide the difference by the mean of the two peaks to obtain the percentage difference between each pair of peaks. If the percentage difference between a peak and the other two peaks exceeds 5%, it means that the peak and the other two peaks do not belong to the same interference source, so the peak is taken as an independent interference frequency. If the percentage difference between two or three peaks does not exceed 5%, it means that they belong to the same interference source, so the mean of these peaks is calculated as the interference frequency. S24: Since the interference frequency is calculated periodically in the above method, after the interference frequency in the current 100 spectrum diagrams is calculated, the spectrum diagrams in the cache queue can be cleared. After clearing the spectrum diagrams in the cache queue, the next week's 100 spectrum diagrams to be calculated are stored for the next calculation. Therefore, while the spectrum diagram of the next period is cached, the system simultaneously performs judgment and filtering operations on the interference frequency calculated in the previous period, thus improving the efficiency of the entire process. S25: Determine whether the fluctuation value between the current interference frequency and the previously stored historical interference frequency exceeds the interference frequency threshold. If it exceeds the interference frequency threshold, the current interference frequency is deemed valid, and proceed to step S3; if it is less than or equal to the interference frequency threshold, the current interference frequency is deemed invalid, and no adjustment of filter parameters is required. The vehicle power environment fluctuates in each acquisition cycle due to factors such as temperature, working environment, and working status. These environmental changes cause small fluctuations in the interference frequency value. However, these small fluctuations do not affect the filter's filtering effect. Therefore, there is no need to store or change the interference frequency to adjust the filter parameters. By setting this fluctuation range as the interference frequency threshold, it is possible to detect in real time whether the current interference frequency needs to enter the subsequent self-learning stage. This fluctuation range can be obtained through product testing; therefore, the interference frequency threshold can be modified according to the device type, test results, and actual needs. In a preferred embodiment, the interference frequency... The interference frequency value is set to ±5%. If the fluctuation between the current interference frequency and the previously stored historical interference frequency (usually the closest historical interference frequency) is within ±5%, it indicates that the variation between the current interference frequency and the previously stored historical frequency is very small. Since the interference frequency and filter parameters have not been updated, the previous filter parameters can be ignored and reused. If the fluctuation between the current interference frequency and the previously stored historical interference frequency exceeds ±5%, it indicates that the difference between the current interference frequency and the previously stored historical interference frequency has exceeded the normal fluctuation range. In this case, the interference source of the current interference frequency is likely different from the historical interference frequency, or the interference frequency may be changing regularly due to equipment aging. In this case, the interference frequency data needs to be updated to adapt to the data changes. Since there can be one or more current interference frequencies, each current interference frequency needs to be compared with the previously stored historical frequency in turn to effectively filter out valid interference frequencies.

[0027] S3: Retrieve all historical interference frequency data stored locally by the camera, compare the current interference frequency with the historical interference frequency using a first threshold and a second threshold. The first threshold is used to determine whether the current interference frequency is within the effective filtering range of the current filter, and the second threshold is used to determine the type of the current interference frequency, distinguishing whether the current interference frequency is a regular anomaly caused by device aging or other abnormal interference frequencies; and dynamically store or change the interference frequency data based on the current interference frequency type to achieve self-learning. like Figure 3 As shown, in some embodiments, step S3 includes the following steps: S31: Retrieve the interference frequency data stored locally by the camera and obtain all historical interference frequencies. The historical interference frequencies refer to the interference frequencies that have been calculated and recorded and have undergone filtering. If this is the first time the device is running, the camera's initial setting value is directly retrieved as the historical interference frequency. The initial setting value is a common interference frequency that is preset based on actual industry experience. The selection of the initial setting value can be adjusted according to actual needs, which will not be elaborated here. S32: Calculate the difference between the current interference frequency and each historical interference frequency, and determine whether the difference between the current interference frequency and each historical interference frequency is greater than the first threshold. If yes, it means that the current interference frequency is not within the effective filtering range of the parameters set by the current filter, and the filter parameters need to be updated. Proceed to step S33. If no, it means that the current interference frequency is within the effective filtering range of the parameters set by the current filter, and there is no need to update the interference frequency data or adjust the filter parameters. The first threshold is calculated based on the effective filtering range of the current filter, which will not be elaborated here. The effective filtering range of the current filter refers to the filtering range obtained by the filter after filtering each previously stored historical parameter.

[0028] S33: Extract the historical interference frequency with the smallest difference from the current interference frequency among all historical interference frequencies, and calculate the difference between the current interference frequency and the historical interference frequency. Determine whether the difference is greater than the second threshold. If yes, it means that the current interference frequency and all historical interference frequencies belong to different interference sources. Therefore, the current interference frequency is stored as a new interference frequency in the interference frequency data. If no, it means that the current interference frequency and the historical interference frequency with the smallest difference belong to the same interference source. However, due to the regular frequency shift caused by equipment aging, the current interference frequency is overwritten by the historical interference frequency with the smallest difference to adapt to the filtering requirements after equipment aging.

[0029] S4: Adjust the filter parameters using the updated interference frequency data and adaptively filter the real-time image data.

[0030] In this process, the filter adjustment parameters calculated using the spectrum acquired in the current period are directly applied to the filtering of the real-time image data corresponding to these spectrums (i.e., filtering from the input source). After filtering, the filtered real-time image data is output to the embedded processor at the back end. This real-time filtering method avoids the situation where the filter is not set to remove the corresponding interference frequencies at the beginning of the device's operation, thus avoiding the problem of interference stripes appearing in the early stages of the image. On the other hand, image information is first extracted by FFT calculation, and after the filter parameters are calculated using this information, they are fed back to the front end for filter setting, thus completing fast filtering at the input source. This way, interference data will not be mixed into the image back end, and large-scale calculations on the real-time image are not required at the image back end.

[0031] In some embodiments, the image filtering operation is performed using a digital filter based on the FIR convolution formula and comb filtering addition and subtraction operations. Through specific coefficient settings, the luminance and chrominance in the composite video signal are accurately separated, and out-of-band noise is suppressed.

[0032] like Figure 4 As shown, this invention also discloses an aging-state-adaptive image interference stripe elimination system for vehicle-mounted cameras, comprising: The acquisition module includes at least one camera for acquiring real-time image data, wherein the real-time image data is a composite video signal; The image processing module is used to convert real-time image data captured by each camera into a spectrogram; The local storage module is used to store all historical interference frequency data; The adaptive learning module performs local self-learning based on the current interference frequency and the historical interference frequencies in the local storage module, and dynamically stores and updates the interference frequency data in the local storage module based on the self-learning results. The filtering module includes filters, and in some embodiments, includes digital filters. The digital filters are based on FIR convolution formulas and comb filtering addition and subtraction operations, and achieve precise separation of luminance and chrominance in composite video signals and suppression of out-of-band noise through specific coefficient settings. The adjustment module is used to configure the filter channel parameters of each channel based on the updated interference frequency data.

[0033] In some embodiments, the vehicle-mounted camera image interference stripe elimination system based on aging state adaptation further includes a caching module, comprising a cache queue for caching Fourier transform spectrograms, real-time recording of the number of cached spectrograms in the cache queue, determining whether the number of spectrograms has reached the target number, and if the target number has been reached, outputting all spectrograms in the cache queue to the adaptive learning module, then clearing the current spectrogram data, and continuing to cache the spectrograms of the next period; in a preferred embodiment, the target number of spectrograms is 100, and when the number of spectrograms in the cache queue reaches 100, the input of spectrograms is temporarily stopped, and the spectrograms of the next period are continued to be cached.

[0034] like Figure 4 As shown, the present invention also discloses a device comprising: at least one camera, an ADC chip, and an embedded processor, wherein when the device has multiple cameras, the interference frequencies of each camera are different, and therefore each needs to perform filtering processing independently. The camera is used to collect real-time image data, which is a composite video signal collected by the camera. The ADC chip includes a digital filter, which performs image filtering based on the FIR convolution formula and comb filtering addition and subtraction operations. The embedded processor receives digital image frames from the ADC chip via the CSI bus, performs image processing such as grayscale conversion, moving window selection, and FFT spectrum analysis to identify the interference frequency that needs to be suppressed. It then performs self-learning by combining historical interference data, comparing the difference between the current interference frequency and the historical interference frequency to determine the type of the current interference frequency. It distinguishes between regular anomalies caused by equipment aging and other abnormal interference frequencies, dynamically updates the interference frequency database, calculates the current filter coefficients based on the learning results, and writes the configuration commands back to the ADC chip via the I²C bus to adjust its internal filter parameters in real time.

[0035] In a preferred embodiment, when the embedded processor calculates the current filter coefficient h int [k] (the filter coefficient h) int [k] is an integer), and the current filter coefficient h int [k] The embedded processor configures the registers inside the ADC chip via the I²C bus, thereby modifying the value of the sharpness control register in real time. The ADC chip, under a 27MHz sampling clock, sets the current filter coefficient h... int [k] is input into the FIR convolution formula, and the digital filter mainly performs filtering operations based on the following FIR convolution formula: ; in: y out [n] represents the filtered output at the nth time step; x[nk] is the sampled value of the input signal at time nk (i.e., the real-time image signal); h int [k] represents the integer tap coefficients of the filter (k=0,1,...,N-1, N orders in total), where the coefficients inside the chip are usually in integer form as fixed-point decimals; This is an arithmetic right shift of S bits, written to the ADC chip by the embedded processor via the I²C bus for gain normalization and to prevent overflow. N is the filter order, which determines the filtering accuracy and computational cost.

[0036] The digital image data y for each real-time image is calculated sequentially using the FIR convolution formula. out [n], thereby achieving precise separation of luminance and chrominance in composite video signals and suppression of out-of-band noise. Finally, the filtered digital image data y is transmitted via the CSI bus. out [n] is transmitted to the embedded processor in real time.

[0037] This invention has many other embodiments, and all technical solutions formed by equivalent transformation or equivalent transformation fall within the protection scope of this invention.

Claims

1. A method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation, characterized in that: Includes the following steps: S1: Acquire real-time image data captured by the camera and convert the real-time image data into a spectrogram; S2: Collect several spectrum graphs in a cycle, extract the highest peak values ​​in all spectrum graphs in each cycle, calculate the current interference frequency, and determine whether the current interference frequency is effective. S3: Retrieve all historical interference frequency data stored locally by the camera, and compare the current interference frequency with the historical interference frequency using a first threshold and a second threshold. The first threshold is used to determine whether the current interference frequency is within the effective filtering range of the current filter, and the second threshold is used to determine the type of the current interference frequency, distinguishing whether the current interference frequency is a regular abnormality caused by device aging or other abnormal interference frequencies. It can dynamically store or change interference frequency data based on the current interference frequency type to achieve self-learning; S4: Adjust the filter parameters using the updated interference frequency data and adaptively filter the real-time image data.

2. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 1, characterized in that: Step S1 includes the following steps: S11: Acquire real-time image data; S12: Perform grayscale processing on each frame of the image; S13: Move the window to reset the detection image area; S14: Perform Fourier transform on the image of the detection region selected by the moving window in each frame to obtain the spectrogram.

3. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 2, characterized in that: In step S13, when setting the detection image region for each frame, the control window moves to a random position and randomly generates the detection image region for the current image.

4. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 2, characterized in that: Step S2 includes the following steps: S21: Store the spectrograms into the cache queue in real time, and determine whether the number of spectrograms in the cache queue has reached the target number. If the target number has been reached, stop storing images and proceed to step S22. S22: Extract the top three highest peaks from all spectrograms in the cache queue; S23: Calculate the current interference frequency based on the three peak values; S24: Clear the spectrogram in the cache queue; S25: Determine whether the fluctuation value between the current interference frequency and the previously stored historical interference frequency exceeds the interference frequency threshold. If it is greater than the interference frequency threshold, the current interference frequency is determined to be valid, and proceed to step S3; if it is less than or equal to the interference frequency threshold, the current interference frequency is determined to be invalid, and no adjustment of filter parameters is required.

5. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 4, characterized in that: Step S23 includes the following steps: Calculate the percentage difference between the three peaks. If the percentage difference between a peak and the other two peaks exceeds 5%, then that peak is taken as an independent interference frequency. If the percentage difference between two or three peaks does not exceed 5%, then the average of these peaks is calculated as the interference frequency.

6. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 4, characterized in that: In step S24, after clearing the spectrograms in the cache queue, new spectrograms are stored for calculation in the next cycle.

7. The method for eliminating interference stripes in vehicle-mounted camera images based on aging state adaptation according to claim 1, characterized in that: Step S3 includes the following steps: S31: Retrieve the interference frequency data stored locally by the camera to obtain all historical interference frequencies; S32: Calculate the difference between the current interference frequency and each historical interference frequency, and determine whether the difference between the current interference frequency and each historical interference frequency is greater than the first threshold; if yes, proceed to step S33; if no, there is no need to update the interference frequency data. S33: Extract the historical interference frequency with the smallest difference from the current interference frequency among all historical interference frequencies, and calculate the difference between the current interference frequency and the historical interference frequency. Determine whether the difference is greater than the second threshold. If yes, store the current interference frequency as a new interference frequency in the interference frequency data. If no, overwrite the historical interference frequency with the one with the smallest difference with the current interference frequency.

8. An image interference stripe elimination system for vehicle-mounted cameras based on aging state adaptation, characterized in that: include: The acquisition module includes at least one camera for acquiring real-time image data. The image processing module is used to convert real-time image data captured by each camera into a spectrogram; The local storage module is used to store all historical interference frequency data; The adaptive learning module performs local self-learning based on the current interference frequency and the historical interference frequencies in the local storage module, and dynamically stores and updates the interference frequency data in the local storage module based on the self-learning results. Filtering module, including filters; The adjustment module is used to configure the filter channel parameters of each channel based on the updated interference frequency data.

9. The vehicle-mounted camera image interference stripe elimination system based on aging state adaptation according to claim 8, characterized in that: It also includes a caching module, which includes a cache queue for caching the spectrograms after Fourier transform. The number of spectrograms cached in the cache queue is recorded in real time. It is determined whether the number of spectrograms has reached the target number. If the target number has been reached, all spectrograms in the cache queue are output to the adaptive learning module. Then, the current spectrogram data is cleared and the spectrograms of the next cycle are cached.

10. The equipment, characterized in that: include: At least one camera, ADC chip, and embedded processor; The camera is used to collect real-time image data; The ADC chip includes a digital filter, which is used to achieve precise separation of luminance and chrominance in composite video signals and suppression of out-of-band noise by setting specific coefficients under a 27MHz sampling clock, and transmits the filtered digital image data to the embedded processor in real time via the CSI bus. The embedded processor receives digital image frames from the ADC chip via the CSI bus, performs grayscale conversion, moving window selection, and FFT spectrum analysis on the image, identifies the interference frequency that needs to be suppressed, and performs self-learning based on historical interference data to dynamically update the interference frequency database. Based on the learning results, it calculates the current filter coefficients and writes the configuration command back to the ADC chip via the I²C bus to adjust its internal filter parameters in real time.

Images