High-definition image optimization method for intelligent connected automobile

Abstract

The invention discloses a high-definition image optimization method for an intelligent networked automobile, relates to the technical field of image optimization, and solves the problems of numerical deviation and blurring of the contour edge of an object caused by gray value overlapping of corresponding areas of front and rear frames of images due to object movement. According to the method, the pixel value difference between adjacent frames is fully utilized, gradient calculation and contour area judgment are combined, the dynamic object can be accurately positioned, compared with a traditional method, misjudgment and missed judgment in a complex environment can be reduced, a reliable basis is provided for subsequent optimization, and the sensing accuracy of an intelligent networked automobile to the surrounding dynamic environment is improved; abnormal contour lines can be quickly identified by comparing gradient features of contour areas of similar dynamic objects in continuous frames; according to the method based on gradient column difference analysis, gray scale deviation generated by pixel point overlapping can be sensitively captured, contour blurring caused by foldover is avoided, and the definition and integrity of the edge of a dynamic object in the image are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of image optimization technology, and in particular to a high-definition image optimization method for intelligent connected vehicles. Background Art

[0002] Against the backdrop of the rapid development of intelligent connected vehicle technology, high-definition images, as the key information carrier for vehicles to perceive the external environment, have a direct impact on the reliability and safety of core functions such as autonomous driving, target recognition, and path planning.

[0003] However, in practical applications, images acquired by intelligent connected vehicles often face many challenges.

[0004] During vehicle driving, dynamic objects are prone to overlapping images when continuous frame images are collected. Due to the movement of the object, the grayscale values ​​of the corresponding areas of the previous and next frame images overlap, resulting in numerical deviations and blurred contours at the edges of the object, which greatly interferes with the accurate extraction of object features in the image. Existing image optimization technologies often have difficulty in accurately identifying dynamic objects and optimizing their contour features in a targeted manner when processing dynamic scenes, resulting in errors in the vehicle's perception of the surrounding environment and increasing the potential risk of traffic accidents.

[0005] In addition, traditional image enhancement and sharpening methods lack in-depth analysis of the characteristics of dynamic objects and cannot effectively deal with the problem of image quality degradation caused by object movement, making it difficult to meet the urgent needs of intelligent connected vehicles for high-definition and accurate image information.

[0006] Therefore, there is an urgent need for a high-definition image optimization method that can effectively identify dynamic objects, optimize their contour features, and improve the overall image quality to ensure the safe and stable operation of intelligent connected vehicles. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides a high-definition image optimization method for intelligent connected vehicles, which solves the problem that the movement of the object causes the grayscale values ​​of the corresponding areas of the previous and next frame images to overlap, resulting in numerical deviations at the edges of the object's contours and blurred contours.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: A high-definition image optimization method for intelligent connected vehicles, comprising the following steps:

[0009] Step 1: Based on the continuous frame images obtained in the specified area, the previous frame image is used as the initial image and the next frame image is used as the image to be optimized;

[0010] The Sobel algorithm is used to confirm the horizontal and vertical gradients associated with different points in the initial image and the image to be optimized, and the following methods are used: To confirm the comprehensive gradient associated with the corresponding pixel point, and mark the pixel point that satisfies: comprehensive gradient > Y1 as a gradient pixel point, where Y1 is the preset value; otherwise, no calibration is performed;

[0011] Based on the calibrated gradient pixel points in the initial image and the image to be optimized, continuous and adjacent gradient pixel points are calibrated as similar pixels. If there is a group of included areas between similar pixels, it is recorded as the contour area. Otherwise, no calibration is performed;

[0012] Based on the calibrated contour areas in the initial image and the image to be optimized, the center point of the contour area is calibrated. From the initial image, it is confirmed whether the center point of the contour area is located within the contour area of ​​the image to be optimized. If so, the two sets of contour areas are recorded as the areas to be compared. Otherwise, no calibration is performed.

[0013] Identify whether the center points of the two groups of areas to be compared are located in the same spatial position in the image. If they are the same, no processing is performed. If they are not the same, the two groups of areas to be compared are synchronously calibrated as the same type of dynamic objects;

[0014] Step 2: The contour area of ​​the dynamic object in the initial image is recorded as the initial area, and the contour area of ​​the dynamic object in the image to be optimized is recorded as the pending area;

[0015] Based on the center point determined in the initial area and several edge pixel points outside the initial area, starting from the center point, a set of plane vectors pointing to the edge pixel points is constructed, and the multiple gradient pixel points associated with the plane vectors are recorded as the initial pixel point column. The integrated gradients associated with the initial pixel point column are sorted according to the direction of the plane vectors to determine the initial integrated gradient column belonging to the plane vector in the initial area;

[0016] Then, based on the calibrated center point in the pending area, the initial point of the plane vector is combined with this center point, and the gradient pixel points associated with the plane vector at the edge contour of the pending area are determined. Simultaneously, based on the orientation of the plane vector, the integrated gradients associated with several gradient pixel points are sorted to generate a pending integrated gradient column belonging to the plane vector in the pending area.

[0017] Confirm the abnormal contour lines of the initial integrated gradient column and the pending integrated gradient column of the same set of plane vectors: confirm the difference between the adjacent integrated gradients of the two integrated gradient columns, and sort the confirmed differences according to the sorting method of the corresponding gradient columns, confirm the initial difference column belonging to the initial integrated gradient column and the pending difference column belonging to the pending integrated gradient column, and each difference is ≥ 0;

[0018] Perform difference processing on the difference between the initial difference column and the difference column to be determined at the same sort position, confirm the sort difference, and then perform average processing on the different sort differences associated with different sort positions to lock the mean to be verified:

[0019] If the mean value to be verified is greater than Y2, several gradient pixel points associated with the undetermined comprehensive gradient column are connected and this connection is calibrated as the abnormal contour line. Several gradient pixel points associated with the initial comprehensive gradient column are connected as the baseline of this abnormal contour line. If the mean value to be verified is less than or equal to Y2, no calibration is performed and Y2 is the preset value.

[0020] Step 3: Confirm the pixels associated with the baseline, determine the different grayscale values ​​associated with different pixels, and determine the minimum and maximum values, and lock the baseline interval [Xmin, Xmax] belonging to this baseline;

[0021] Then use the same processing method to lock the associated interval [Gmin, Gmax] belonging to this abnormal contour line, using: a = (Gmax-Gmin) ÷ (Xmax-Xmin), b = Xmin-a×Gmin, and lock the two constants a and b;

[0022] The grayscale values ​​associated with different pixel points within the abnormal contour are recorded as H i , where i represents different pixels, using T i =a×H i +b confirms the grayscale value T associated with the corresponding pixel point to be adjusted i , and based on the gray value T to be adjusted i The pixel value of this pixel point is adjusted and optimized. Similarly, the pixel values ​​associated with different pixel points within the abnormal contour line are adjusted and optimized. If T i >255, then adjust and optimize according to 255. If T i <0, then adjust and optimize according to 0;

[0023] Step 4: Based on the different undetermined comprehensive gradient columns associated with each different plane vector, the gradient pixel points associated with the undetermined comprehensive gradient columns are confirmed, the comprehensive gradients associated with the gradient pixel points are reconfirmed, and based on the different comprehensive gradients associated with several gradient pixel points, the maximum value is selected, and the gradient pixel point associated with the maximum value is marked as the point to be sharpened, and several points to be sharpened existing in the edge contour of this contour area are confirmed in turn;

[0024] The pixel values ​​associated with several pixel points within the edge contour are averaged, the associated mean J is locked, and the grayscale values ​​associated with different points to be sharpened are calibrated as H. q , where q represents different points to be sharpened, using: TZq =(H q -J)+H q Lock the grayscale value TZ corresponding to the point to be sharpened q , and readjust the grayscale value of the sharpened point. If TZ q >255, then the sharpening adjustment will be made according to the value 255.

[0025] The present invention provides a high-definition image optimization method for intelligent connected vehicles. Compared with the existing technology, it has the following advantages:

[0026] This method fully utilizes the difference in pixel values ​​between adjacent frames, combines gradient calculation with contour area determination, and can accurately locate dynamic objects. Compared with traditional methods, it can reduce misjudgments and missed judgments in complex environments, provide a reliable foundation for subsequent optimization, and improve the accuracy of intelligent connected vehicles' perception of the surrounding dynamic environment.

[0027] By comparing the gradient features of the contour areas of similar dynamic objects in consecutive frames, abnormal contour lines can be quickly identified. This method, based on gradient column difference analysis, can keenly capture grayscale deviations caused by pixel overlap, avoiding contour blur caused by overlapping images, significantly improving the clarity and integrity of the edges of dynamic objects in the image, and enabling the vehicle to more accurately recognize the object's shape.

[0028] Based on the grayscale feature relationship between the baseline and the abnormal contour line, a linear transformation model is established to adjust the grayscale value and achieve adaptive optimization. This strategy can effectively correct the grayscale deviation of the abnormal contour line, making the image grayscale distribution more consistent with the actual scene, enhancing the image layering and visual effects, and providing better image information for the vehicle vision system.

[0029] By identifying the points to be sharpened and re-adjusting their grayscale, key features of dynamic object contours are highlighted, enhancing the contrast and clarity of outline edges. This sharpening process not only improves the visual quality of the image but also enables the vehicle to more accurately identify object boundaries in tasks such as target detection and path planning, improving the reliability and safety of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Schematic diagram of the process of the present invention;

[0031] Figure 2 Schematic diagram for determining the corresponding plane vector of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] See also Figure 1 , this application provides a high-definition image optimization method for intelligent connected vehicles, comprising the following steps:

[0034] Step 1: Based on the acquisition instruction, continuous frame images of the specified area are generated, and the continuous frame images are processed frame by frame, dynamic objects are identified from adjacent frames, and marked synchronously. The acquisition instruction is issued by the vehicle computer or the driver. Specifically, when it is necessary to acquire images of the specified area, a group of images is acquired first, and then another group of images is acquired based on the acquired images. The two groups of images are continuous frames. The continuous frames may be separated by one second or less. The specific value is determined by the operator based on experience and is generally not too long. The reason for confirming two groups of frames is to better identify dynamic objects and facilitate subsequent image optimization to achieve better image optimization processing effects. The latter group of continuous frame images is the optimized image.

[0035] Among them, the specific method of identifying dynamic objects is:

[0036] Based on the continuous frame images obtained in the specified area, the previous frame image is used as the initial image and the next frame image is used as the image to be optimized;

[0037] The Sobel algorithm is used to confirm the horizontal and vertical gradients associated with different points in the initial image and the image to be optimized (the horizontal and vertical gradients are confirmed based on the difference in pixel values ​​between corresponding pixels. The corresponding image is first grayscaled to confirm the grayscale image. Then, based on the pixel features in the grayscale image, the grayscale values ​​associated with the corresponding points are confirmed. Based on the grayscale value difference of the nine-square grid, the horizontal and vertical gradients associated with the corresponding central pixel are determined. The gradient values ​​of the outermost pixels of the image are not specifically confirmed). Based on the confirmed horizontal and vertical gradients, the following method is used: To confirm the comprehensive gradient associated with the corresponding pixel point, and calibrate the pixel point that satisfies: comprehensive gradient > Y1 as a gradient pixel point, where Y1 is a preset value, and its specific value is determined by the operator based on experience. Otherwise, no calibration is performed. Several gradient pixel points in the initial image and the image to be optimized are calibrated in turn;

[0038] Based on the calibrated gradient pixels in the initial image and the image to be optimized, continuous and adjacent gradient pixels are calibrated as similar pixels. If there is a group of included areas between similar pixels, it is recorded as the contour area (here, the circle of points around the contour area are all gradient pixels, that is, a closed-loop gradient contour is formed). Otherwise, no calibration is performed (if a group of encircling circles is not formed, there may be a texture area between the gradient points. Such areas are all gradient pixels, but no encirclement is formed. Under normal circumstances, when there is a significant grayscale difference in an object, the points associated with its edge contour are all gradient pixels);

[0039] Based on the calibrated contour areas in the initial image and the image to be optimized, the center point of the contour area is calibrated (combined with the two-dimensional coordinate system, the two-dimensional coordinates associated with different pixel points in the contour area can be confirmed, and then several two-dimensional coordinates are averaged to confirm the average coordinates, so that the center point can be calibrated). From the initial image, confirm whether the center point of the contour area is located in the contour area of ​​the image to be optimized. If so, the two groups of contour areas are recorded as the areas to be compared. Otherwise, no calibration is performed;

[0040] Identify whether the center points of the two sets of compared areas are located in the same spatial position in the image. If they are the same, no processing is performed. If they are not the same, the two sets of compared areas are synchronously calibrated as the same type of dynamic objects. Specifically, the time difference between the two previous and next frame images is very short when they are confirmed. Therefore, when the dynamic object moves, the center point generated will not deviate from the specific range of the next frame image. In addition, the acquisition angles and various parameters of the previous and next frame images are consistent. If their spatial positions are the same, it means that the object associated with such areas belongs to a static object. Otherwise, it belongs to a dynamic object.

[0041] When verifying spatial position, the time between consecutive frames is short, so there is no need to consider vehicle speed. In addition, when confirming spatial position, the system belongs to a range area. That is, when two points are synchronously within the specified range of the corresponding image, it means that the spatial position is the same. The range area is determined by the length and width of the image, generally a range size of (0.005 length × 0.005 width);

[0042] Step 2: Based on the similar dynamic objects marked in the continuous frame images, the contour features of the corresponding dynamic objects in the previous frame image are used as the benchmark to identify the abnormal contour lines in the next frame image. Subsequently, the grayscale adjustment and optimization of the existing abnormal contour lines are performed through step 3. This is a superposition process. When the object moves, because of the image acquisition process, the grayscale values ​​of the area where the object moves and the objects originally belonging to this area will overlap. This situation will cause the contour edge of the corresponding object to have a numerical deviation, which will easily cause the contour to be unclear. Therefore, in order to avoid this situation, the abnormal points on the contour edge of the dynamic object are confirmed and subsequently optimized to achieve better image optimization effects. The specific method of identifying abnormal contour lines is as follows:

[0043] Based on the similar dynamic objects marked in the continuous frame images, the contour area of ​​the dynamic object in the initial image is recorded as the initial area, and the contour area of ​​the dynamic object in the image to be optimized is recorded as the pending area;

[0044] Based on the center point determined in the initial area and several edge pixel points outside the initial area, starting from the center point, a set of plane vectors facing the edge pixel points is constructed, and the multiple gradient pixel points associated with this plane vector are recorded as the initial pixel point column, and the comprehensive gradient associated with the initial pixel point column is sorted according to the direction of this plane vector, and the initial comprehensive gradient column belonging to this plane vector in the initial area is confirmed (different edge pixel points are associated with different plane vectors, and different plane vectors can be associated with different initial comprehensive gradient columns. Why are there multiple gradient pixel points here? First, the edge contour of an object does not only have one gradient pixel point, but an area generated by several gradient pixel points. This type of edge contour is not a line, but an area, so there are several gradient pixel points in this area), combined with Figure 2 , determine the center point in the corresponding contour area, and then confirm the corresponding plane vector based on the center point and the edge pixel points of the corresponding contour area. Then, based on the corresponding intersection points generated between the plane vector and the corresponding edge contour, the intersection points are the corresponding gradient pixel points. After the gradient pixel points are sorted, they become the associated comprehensive gradient column, and the specific direction of the sorting is the direction of the corresponding plane vector;

[0045] Then, based on the calibrated center point in the pending area, the initial point of the plane vector is combined with this center point, and the gradient pixel points associated with the plane vector at the edge contour of the pending area are determined. Simultaneously, based on the orientation of the plane vector, the integrated gradients associated with several gradient pixel points are sorted to generate a pending integrated gradient column belonging to the plane vector in the pending area.

[0046] The initial integrated gradient column and the to-be-determined integrated gradient column of the same set of plane vectors are used to confirm the abnormal contour lines: the difference between the adjacent integrated gradients of the two integrated gradient columns is confirmed, and the confirmed differences are sorted according to the sorting method of the corresponding gradient columns, and the initial difference column belonging to the initial integrated gradient column and the to-be-determined difference column belonging to the to-be-determined integrated gradient column are confirmed, and each difference is ≥ 0 (the proposed initial integrated gradient column is {Z1, Z2, Z3, Z4, ..., Zn}, and the associated difference column is {|Z1-Z2|, |Z2-Z3|, |Z3-Z4|, ...}). For the to-be-determined integrated gradient column, the difference processing method is also the same. When the dynamic object moves, if there is no pixel overlap, the two gradient difference columns should be relatively consistent and will not change significantly. However, if there is a large difference, it means that there is pixel overlap at the corresponding contour edge in the image to be optimized, and optimization processing is required;

[0047] Perform difference processing on the difference between the initial difference column and the difference column to be determined at the same sorting position to confirm the sorting difference (the difference processing method here is: |difference of initial difference column - difference of difference column to be determined|, and the sorting difference here is also ≥ 0). Then perform average processing on the different sorting differences associated with different sorting positions to lock the average to be verified:

[0048] If the mean value to be checked is greater than Y2, then a line is connected between several gradient pixel points associated with the undetermined comprehensive gradient column, and this line is marked as the abnormal contour line. In addition, a line is connected between several gradient pixel points associated with the initial comprehensive gradient column, and this line is used as the baseline of the abnormal contour line.

[0049] Otherwise, no calibration is performed and Y2 is the preset value. The specific value is determined by the operator based on experience.

[0050] Step 3: Based on the calibrated abnormal contour line and the associated baseline, according to the pixel features of the baseline, the grayscale values ​​associated with different pixels of the abnormal contour line are synchronously optimized and adjusted. The specific method of adjustment is as follows:

[0051] Confirm the pixels associated with the baseline, determine the different grayscale values ​​associated with different pixels, and determine the minimum and maximum values, and lock the baseline interval [Xmin, Xmax] belonging to this baseline;

[0052] Then use the same processing method to lock the associated interval [Gmin, Gmax] belonging to this abnormal contour line, using: a = (Gmax-Gmin) ÷ (Xmax-Xmin), b = Xmin-a×Gmin, and lock the two constants a and b;

[0053] The grayscale values ​​associated with different pixel points within the abnormal contour are recorded as H i , where i represents different pixels, using T i =a×H i +b confirms the grayscale value to be adjusted associated with the corresponding pixel point, and adjusts and optimizes the pixel value of this pixel point based on the grayscale value to be adjusted. Similarly, the pixel values ​​associated with different pixel points within the abnormal contour line are adjusted and optimized. If T i >255, then adjust and optimize according to 255. If T i <0, then adjust and optimize according to 0 to complete the contour optimization process of the image to be optimized;

[0054] Step 4: For the image to be optimized after contour optimization and adjustment, based on the contour features associated with the contour area of ​​the dynamic object in the image to be optimized, identify the points to be sharpened, and readjust the grayscale values ​​of the points to be sharpened to complete the sharpening process. The specific method for identifying the points to be sharpened is:

[0055] Based on the different undetermined integrated gradient columns associated with each different plane vector, the gradient pixel points associated with the undetermined integrated gradient column are confirmed (because the corresponding grayscale value has been optimized and adjusted, the gradient needs to be reconfirmed), the integrated gradient associated with the gradient pixel points is reconfirmed, and based on the different integrated gradients associated with several gradient pixel points (the several pixel points here are several pixel points associated with a comprehensive gradient column, not all), the maximum value is selected, the gradient pixel point associated with the maximum value is marked as the point to be sharpened, and several points to be sharpened existing in the edge contour of this contour area are confirmed in turn;

[0056] The pixel values ​​associated with several pixel points within the edge contour are averaged, the associated mean J is locked, and the grayscale values ​​associated with different points to be sharpened are calibrated as H. q , where q represents different points to be sharpened, using: TZ q =(H q -J)+H q Lock the grayscale value TZ corresponding to the point to be sharpened q , and readjust the grayscale value of the sharpening point to complete the sharpening process. If TZ q >255, then the sharpening adjustment is performed according to the value 255;

[0057] It also includes: re-converting the image to be optimized after sharpening into an RGB image and displaying it. The conversion process is opposite to the corresponding grayscale processing process, which is more common in the existing technology. That is, based on the adjusted grayscale value, the associated RGB value is adjusted year-on-year for re-display, so as to achieve a stronger image display effect.

[0058] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0059] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A high-definition image optimization method for intelligent connected vehicles, characterized in that: The following steps are involved: Step 1: Based on the acquisition instruction, generate continuous frame images of the specified area, perform frame processing on the continuous frame images, identify dynamic objects from adjacent frames, and mark them synchronously; Step 2: Based on the similar dynamic objects marked in the continuous frame images, the contour features of the corresponding dynamic objects in the previous frame image are used as a reference to identify the abnormal contour lines in the next frame image; Step 3: Based on the calibrated abnormal contour line and the associated baseline, according to the pixel features of the baseline, the grayscale values ​​associated with different pixels of the abnormal contour line are synchronously optimized and adjusted; Step 4: For the image to be optimized after the contour optimization adjustment is completed, based on the contour features associated with the contour area of ​​the dynamic object in the image to be optimized, the points to be sharpened are identified, and the grayscale values ​​of the points to be sharpened are readjusted to complete the sharpening process.

2. The high-definition image optimization method for intelligent connected vehicles according to claim 1, characterized in that: In step 1, the specific method of identifying the dynamic object is: Based on the continuous frame images obtained in the specified area, the previous frame image is used as the initial image and the next frame image is used as the image to be optimized; The Sobel algorithm is used to confirm the horizontal and vertical gradients associated with different points in the initial image and the image to be optimized, and the following methods are used: To confirm the comprehensive gradient associated with the corresponding pixel point, and mark the pixel point that satisfies: comprehensive gradient > Y1 as a gradient pixel point, where Y1 is the preset value; otherwise, no calibration is performed; Based on the calibrated gradient pixel points in the initial image and the image to be optimized, continuous and adjacent gradient pixel points are calibrated as similar pixels. If there is a group of included areas between similar pixels, it is recorded as the contour area. Otherwise, no calibration is performed; Based on the calibrated contour areas in the initial image and the image to be optimized, the center point of the contour area is calibrated. From the initial image, it is confirmed whether the center point of the contour area is located within the contour area of ​​the image to be optimized. If so, the two sets of contour areas are recorded as the areas to be compared. Otherwise, no calibration is performed. Identify whether the center points of the two groups of areas to be compared are located in the same spatial position in the image. If they are the same, no processing is performed. If they are not the same, the two groups of areas to be compared are synchronously calibrated as the same type of dynamic objects.

3. The high-definition image optimization method for intelligent connected vehicles according to claim 2, characterized in that: In step 2, the specific method of identifying abnormal contour lines is: The contour area of ​​the dynamic object in the initial image is recorded as the initial area, and the contour area of ​​the dynamic object in the image to be optimized is recorded as the pending area; Based on the center point determined in the initial area and several edge pixel points outside the initial area, starting from the center point, a set of plane vectors pointing to the edge pixel points is constructed, and the multiple gradient pixel points associated with the plane vectors are recorded as the initial pixel point column. The integrated gradients associated with the initial pixel point column are sorted according to the direction of the plane vectors to determine the initial integrated gradient column belonging to the plane vector in the initial area; Then, based on the calibrated center point in the pending area, the initial point of the plane vector is combined with this center point, and the gradient pixel points associated with the plane vector at the edge contour of the pending area are determined. Simultaneously, based on the orientation of the plane vector, the integrated gradients associated with several gradient pixel points are sorted to generate a pending integrated gradient column belonging to the plane vector in the pending area. Confirm the abnormal contour lines of the initial integrated gradient column and the pending integrated gradient column of the same set of plane vectors: confirm the difference between the adjacent integrated gradients of the two integrated gradient columns, and sort the confirmed differences according to the sorting method of the corresponding gradient columns, confirm the initial difference column belonging to the initial integrated gradient column and the pending difference column belonging to the pending integrated gradient column, and each difference is ≥ 0; Perform difference processing on the difference between the initial difference column and the difference column to be determined at the same sort position, confirm the sort difference, and then perform average processing on the different sort differences associated with different sort positions to lock the mean to be verified: If the mean value to be checked is greater than Y2, several gradient pixel points associated with the undetermined comprehensive gradient column are connected, and this connection is marked as the abnormal contour line. Several gradient pixel points associated with the initial comprehensive gradient column are connected as the baseline of this abnormal contour line.

4. The high-definition image optimization method for intelligent connected vehicles according to claim 3, characterized in that: If the mean value to be checked is ≤ Y2, no calibration is performed and Y2 is the preset value.

5. The high-definition image optimization method for intelligent connected vehicles according to claim 1, characterized in that: In step 3, the specific method of synchronously optimizing and adjusting the grayscale values ​​associated with different pixel points is: Confirm the pixels associated with the baseline, determine the different grayscale values ​​associated with different pixels, and determine the minimum and maximum values, and lock the baseline interval [Xmin, Xmax] belonging to this baseline; Then use the same processing method to lock the associated interval [Gmin, Gmax] belonging to this abnormal contour line, using: a = (Gmax-Gmin) ÷ (Xmax-Xmin), b = Xmin-a×Gmin, and lock the two constants a and b; The grayscale values ​​associated with different pixel points within the abnormal contour are recorded as H i , where i represents different pixels, using T i =a×H i +b confirms the grayscale value T associated with the corresponding pixel point to be adjusted i , and based on the gray value T to be adjusted i The pixel value of this pixel point is adjusted and optimized, and similarly, the pixel values ​​associated with different pixel points within the abnormal contour line are adjusted and optimized.

6. The high-definition image optimization method for intelligent connected vehicles according to claim 5, characterized in that: If T i >255, then adjust and optimize according to 255. If T i If <0, then adjust and optimize according to 0.

7. The high-definition image optimization method for intelligent connected vehicles according to claim 5, characterized in that: In step 4, the specific method for confirming the sharpening point is: Based on the different pending comprehensive gradient columns associated with each different plane vector, the gradient pixel points associated with the pending comprehensive gradient columns are confirmed, the comprehensive gradients associated with the gradient pixel points are reconfirmed, and based on the different comprehensive gradients associated with several gradient pixel points, the maximum value is selected, the gradient pixel point associated with the maximum value is marked as the point to be sharpened, and several points to be sharpened existing in the edge contour of this contour area are confirmed in turn.

8. The high-definition image optimization method for intelligent connected vehicles according to claim 7, characterized in that: In step 4, the specific method of re-adjusting the grayscale value of the point to be sharpened is: The pixel values ​​associated with several pixel points within the edge contour are averaged, the associated mean J is locked, and the grayscale values ​​associated with different points to be sharpened are calibrated as H. q , where q represents different points to be sharpened, using: TZ q =(H q -J)+H q Lock the grayscale value TZ corresponding to the point to be sharpened q , and readjust the grayscale value of the sharpened point. If TZ q >255, then the sharpening adjustment will be made according to the value 255.

Images