Automobile 360 panoramic image stitching quality analysis method based on machine vision
Through machine vision technology, real-time acquisition of environmental images and driving speeds, establishing time series, identifying splicing defects, and analyzing splicing stability, solving the problems of inaccurate static evaluation and neglecting splicing time in the existing technology, and achieving the splicing quality of 360 panoramic images Dynamic assessment and stability improvement.
Patent Information
- Application Number
- CN202411172954.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-26
AI Technical Summary
The existing 360 panoramic image splicing quality evaluation method is mainly static evaluation, which cannot truly reflect the actual performance of splicing quality during driving of the car, and lack of attention to the splicing processing time, which may lead to inaccurate evaluation results and bottlenecks in system performance, affecting drivers. Real-time judgment and security.
Using a machine vision-based method, the environment image and driving speed are collected in real time through panoramic imaging equipment, speed sensors and processors, to form a time series, establish the corresponding relationship between driving speed and splicing time, and to identify splicing ghosting, misalignment, loss and gaps. The stability and quality stability of splicing processing were analyzed and dynamic evaluation was performed.
The real and comprehensive dynamic evaluation of the quality of the panoramic image of the automobile 360 is achieved, which improves the reliability and practical value of the evaluation results, and ensures the stability and safety of the splicing processing at different speeds.
Smart Images

Figure CN118691954B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of image stitching, and in particular relates to a method for analyzing the stitching quality of automobile 360-degree panoramic images based on machine vision. Background Art
[0002] The automotive 360-degree panoramic imaging system is a technology that integrates multiple cameras. It collects images of the vehicle environment from different angles and stitches them into a 360-degree panoramic image, providing a comprehensive view of the vehicle's surroundings, helping the driver see blind spots that traditional lenses cannot cover, and reducing the risk of collisions caused by limited field of view. However, when the stitching quality is poor, it may cause gaps, misalignment, loss or ghosting in the panoramic image, which will cause the driver to misjudge the environment around the vehicle. It can be seen that the stitching quality of the 360-degree panoramic imaging system has a direct impact on the safety provided by the panoramic imaging system, and it is particularly necessary to analyze and evaluate the stitching quality of the 360-degree panoramic image.
[0003] Although the Chinese invention patent with the authorization announcement number CN113194309B provides a 360-degree panoramic view device imaging quality assessment method, including black and white checkerboard, ambient lighting system, vehicle, 360-degree panoramic view device and imaging quality assessment system, by turning on the ambient lighting system, starting the vehicle and 360-degree panoramic view device, the panoramic image is taken by the surround view device and fused into a panoramic stitching image and sent to the imaging quality assessment system, the system performs stitching ghost, stitching misalignment, loss image and stitching gap recognition, and forms the black and white checkerboard information of these positions, and then converts it into the value of the actual panoramic stitching image to generate a test report. However, the method's assessment of the stitching quality of 360 panoramic images belongs to a static assessment, ignoring that the actual formation environment of 360 panoramic image stitching is during the driving of the car, and the driving of the car is a dynamic process accompanied by changes in driving speed. Relying only on static assessment cannot truly reflect the actual performance of the stitching quality during the driving of the car, which may cause the assessment results to be inconsistent with the performance in actual applications, greatly affecting the reliability and practical value of the assessment results.
[0004] In addition, the above method focuses on the image itself when evaluating the quality of 360-degree panoramic image stitching, and lacks attention to the stitching processing time, which may mask the performance bottleneck of the system. If the stitching algorithm or hardware processing power is insufficient, it will cause processing delays and affect the driver's real-time judgment of the environment. Especially when driving at high speeds, delays in stitching processing time can easily increase the risk of accidents. Summary of the invention
[0005] In view of this, the present invention aims to propose a method for analyzing the stitching quality of a 360-degree panoramic image of a car based on machine vision, and to analyze and evaluate the stitching quality of the 360-degree panoramic image of the car from a dynamic dimension according to the change in the driving speed of the car.
[0006] The purpose of the present invention can be achieved through the following technical solutions: a method for analyzing the quality of 360-degree panoramic image stitching of a car based on machine vision, comprising a panoramic camera device, a speed sensor, an ambient lighting terminal and a processor, wherein the panoramic camera device is composed of cameras arranged in various directions, and the ambient lighting system is composed of multiple light sources, which evenly cover the field of view of the panoramic camera device, and specifically includes the following steps: S1, using the panoramic camera device installed in the car to collect environmental images of each arrangement direction in real time during the driving process of the car, and transmit them to the processor, and the processor stitches the transmitted environmental images of each arrangement direction, and records the stitching completion time at the same time to form a panoramic stitching image time series.
[0007] S2. Use the speed sensor installed in the car to detect the driving speed in real time during the driving process of the car, and record the driving speed detection time to form a driving speed time series.
[0008] S3. Mapping the driving speed time series with the panoramic stitching image time series, thereby establishing a corresponding relationship between the driving speed and the stitching duration.
[0009] S4. Perform stitching ghost, stitching misalignment, loss image, and stitching gap identification on the panoramic stitching image time series respectively to obtain stitching quality parameters of each panoramic stitching image, specifically including stitching ghost parameters, stitching misalignment parameters, loss image parameters, and stitching gap parameters.
[0010] S5. Based on the driving speed time series and the corresponding relationship between the driving speed and the splicing duration, a correlation comparison is performed between the driving speed and the splicing duration, thereby analyzing the splicing processing stability.
[0011] S6. Form a correlation curve between the stitching quality parameters of the panoramic stitching image time series and the driving speed time series, thereby analyzing the stitching quality stability.
[0012] S7. Determine the advantageous splicing quality parameter and the steady-state driving speed based on the formed correlation curve.
[0013] In combination with all the above-mentioned technical solutions, the positive effect possessed by the present invention is as follows: the present invention forms a panoramic stitching image time series and a driving speed time series by synchronously performing panoramic image acquisition and driving speed detection during the driving of the car, thereby performing a correlation comparison between the stitching time and the stitching speed and a correlation comparison between the stitching quality parameter and the driving speed, thereby realizing a dynamic evaluation of the stitching quality of the car's 360-degree panoramic image based on the driving speed combined with the stitching time from the perspective of the stitching processing stability and the stitching quality stability, which can truly and comprehensively reflect the actual performance of the stitching quality during the driving of the car, and greatly improve the reliability and practical value of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.
[0015] Figure 1 The present invention is a flowchart of the steps for implementing the method.
[0016] Figure 2 Schematic diagram of the equipment connection used in the present invention.
[0017] Figure 3 It is a schematic diagram of the association curve division in the present invention. DETAILED DESCRIPTION
[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.
[0019] See also Figure 1 As shown, the present invention proposes a method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision, including a panoramic camera device, a speed sensor, an ambient lighting terminal and a processor, wherein the panoramic camera device is composed of cameras arranged in various directions, and the ambient lighting terminal is composed of multiple light sources, which evenly cover the field of view of the panoramic camera device.
[0020] In the specific application of the present invention, the panoramic camera device is used to collect the environmental image around the vehicle body, the speed sensor is used to collect the driving speed of the car, and the environmental lighting system is used to set each light source at the layout position of each camera to cover the field of view of the panoramic camera device, and provide suitable and balanced lighting conditions. For example, the provided environmental light illumination range can be between 500 lux and 100,000 lux, which helps to ensure that the images captured by each camera have consistent brightness and contrast, which helps to reduce the stitching defects caused by lighting differences. At the same time, the consistency of lighting conditions is ensured so that the stitching quality of the panoramic image can be evaluated under controllable conditions, and the factors of lighting changes can be eliminated, which helps to more accurately evaluate the performance of the 360 panoramic image at different speeds and ensure that the stitching quality is not affected by lighting changes. The processor is used to stitch the environmental images collected by the panoramic camera device to form a 360 panoramic image.
[0021] The hardware device connection involved in the present invention is as follows Figure 2 shown.
[0022] The present invention specifically includes the following steps: S1, using a panoramic camera device installed in a car to collect environmental images of various layout orientations in real time during the driving process of the car, and transmit them to a processor, which stitches the transmitted environmental images of various layout orientations and records the stitching completion time to form a panoramic stitching image time series.
[0023] S2. Use the speed sensor installed in the car to detect the driving speed in real time during the driving process of the car, and record the driving speed detection time to form a driving speed time series.
[0024] It should be added that the panoramic camera equipment and speed sensor maintain a consistent acquisition frequency when collecting environmental images and driving speed. This ensures the synchronization of image data and speed data in time. This means that the image data captured at each time point can accurately correspond to the speed data at that time, avoiding time deviation between data, and also facilitating the subsequent mapping of driving speed time series with panoramic stitching image time series.
[0025] S3. Map the driving speed time series and the panoramic stitching image time series to establish a corresponding relationship between the driving speed and the stitching duration. Specifically, the corresponding relationship is established as follows: the panoramic stitching images in the panoramic stitching image time series are numbered according to the arrangement order, and similarly, the driving speeds in the driving speed time series are numbered.
[0026] According to the one-to-one correspondence principle, the panoramic stitching images and driving speeds with corresponding numbers are extracted from the panoramic stitching image time series and the driving speed time series to form a mapping group.
[0027] The stitching completion time and the driving speed detection time of the panoramic stitching images in each mapping group are extracted, thereby calculating the stitching duration corresponding to the driving speed in each mapping group.
[0028] A corresponding relationship between the driving speed and the splicing duration is established based on the splicing duration corresponding to the driving speed in each mapping group.
[0029] It should be understood that although panoramic cameras and speed sensors maintain a consistent acquisition frequency when collecting environmental images and driving speeds, image acquisition is usually carried out in real time, while the stitching process may be processed offline, which means that there may be time inconsistencies between image acquisition and stitching operations. In addition, image stitching involves complex calculations and processing steps, such as feature matching, image fusion, etc. These steps take time. Therefore, even if the acquisition frequency is consistent, the processing and stitching algorithms may introduce delays. Furthermore, the image data may need to be cached before processing, thereby introducing time intervals.
[0030] S4. Perform stitching ghost, stitching misalignment, loss image, and stitching gap identification on the panoramic stitching image time series respectively to obtain stitching quality parameters of each panoramic stitching image, specifically including stitching ghost parameters, stitching misalignment parameters, loss image parameters, and stitching gap parameters.
[0031] In a preferred implementation of the above solution, the stitching ghost recognition uses an edge detection algorithm to identify repeated edges in the image, thereby finding ghosts, and then marking the ghost areas in the panoramic stitching image.
[0032] Stitching misalignment recognition can detect misalignment at the stitching point through a feature point matching algorithm. If there is a deviation in the matching of feature points, there may be a misalignment, and the misaligned area will be marked in the panoramic stitching image.
[0033] As another example, the splicing misalignment identification may generate an error map of the splicing area, and use the error map to identify the splicing misalignment, wherein a high error area generally indicates a misalignment.
[0034] As another example, stitching misalignment identification can analyze whether the geometric transformation (such as translation, rotation) between images is consistent with expectations. If the transformation is inconsistent, misalignment may exist.
[0035] Loss image recognition can use analytical technology to detect occlusion or missing parts of the stitching area, thereby identifying the existence of image loss and marking the loss area in the panoramic stitching image.
[0036] Stitching gap identification can use image processing algorithms (such as edge detection and contour analysis) to detect the gaps at the image stitching. The gaps usually appear as discontinuities or color changes at the image edges, and mark the gap areas in the panoramic stitching image.
[0037] As another example, the stitching seam identification may be performed by comparing the lighting consistency of the stitching area. Inconsistent lighting may indicate a stitching seam. By comparing the lighting and color of the stitching area, the seam may be identified.
[0038] Applied to the above embodiments, the stitching ghosting parameters may be the ghosting area and the ghosting amplitude, wherein the ghosting amplitude is the pixel brightness difference calculated in the ghosting area; the stitching misalignment parameters may be the misalignment area and the error amplitude, wherein the error amplitude is the amount of misalignment in image stitching, which is usually expressed by calculating the displacement between images; the loss image parameter may be the degree of loss, which may be expressed by calculating the proportion or area of the loss area; and the stitching gap parameter may be the gap width.
[0039] The present invention performs stitching quality analysis on 360-degree panoramic images from four aspects: stitching ghost, stitching misalignment, image loss, and stitching gaps. This makes the stitching quality analysis more comprehensive and can more accurately reflect the overall effect of image stitching. Each aspect can capture different types of quality problems, thereby providing a more complete quality assessment.
[0040] S5. Based on the driving speed time series and the corresponding relationship between the driving speed and the splicing duration, a correlation comparison is performed between the driving speed and the splicing duration, thereby analyzing the splicing processing stability.
[0041] Preferably, the correlation and comparison process between the driving speed and the splicing time is as follows: according to the driving speed time series, a number of points are marked in a coordinate system constructed with time as the horizontal coordinate and the driving speed as the vertical coordinate to form a driving speed change curve.
[0042] In the formed driving speed variation curve, the splicing time is used as the ordinate, and the splicing time corresponding to each driving speed is marked according to the corresponding relationship between the driving speed and the splicing time to form a splicing time variation curve.
[0043] It should be added that, in a coordinate system, data of a horizontal coordinate and multiple vertical coordinates are drawn. This is usually used to represent the relationship between multiple different vertical coordinate values at a specific horizontal coordinate value. In the driving speed change curve constructed based on the driving speed time series, since there is a corresponding relationship between the driving speed and the splicing duration, a splicing duration change curve can be formed based on the construction of the driving speed change curve.
[0044] Further preferably, the stability analysis of the splicing process is as follows: inflection points are marked on the driving speed change curve, and the driving speed corresponding to each inflection point is marked.
[0045] The inflection point mentioned above refers to the point where the curvature of the curve changes.
[0046] Based on the driving speed corresponding to each inflection point, the reference points are marked on the splicing duration change curve, and the tangent slope of each reference point is obtained as the splicing duration change rate of each reference point.
[0047] Based on the change rate of the splicing duration of each inflection point corresponding to the control point, the analytical formula Get the stability of splicing processing , where Indicates The change rate of the splicing duration of the inflection point corresponding to the control point, Indicates the inflection point number, In this formula, when the driving speed changes, the smaller the change in splicing time is, the greater the stability of the splicing process is.
[0048] It should be emphasized that the present invention focuses on the stability of the splicing processing time in the scenario of changing driving speed, rather than the processing delay. This is because the stability analysis ensures that the fluctuation of the splicing processing time at different speeds is minimal, which helps to ensure the consistency of system performance and provide reliable splicing quality even in a rapidly changing environment.
[0049] It should be understood that when performing the splicing processing stability analysis, the point where the driving speed changes is selected on the driving speed change curve, and then the splicing time change rate corresponding to the point is obtained, thereby performing the splicing processing stability analysis. This analysis method directly considers the changing factors of the driving speed, and can identify the key factors affecting the stability of the splicing processing in a dynamic environment, thereby providing an accurate evaluation of the performance of the splicing system under dynamic conditions, and can more realistically and flexibly reflect the performance of the splicing processing in a dynamic environment. This method not only provides a more detailed and comprehensive performance evaluation, but also provides a solid foundation for subsequent algorithm optimization and system improvement, thereby significantly improving the efficiency and practical application value of the splicing processing stability analysis.
[0050] S6. Form a correlation curve between the stitching quality parameters of the panoramic stitching image time series and the driving speed time series, thereby analyzing the stitching quality stability.
[0051] In the specific implementation of the above solution, the correlation curve is formed as follows: the driving speeds existing in the driving speed time series are arranged in order from small to large.
[0052] According to the mapping group composed of the panoramic stitching image and the driving speed, the stitching ghost parameters, stitching misalignment parameters, loss image parameters and stitching gap parameters of the panoramic stitching image corresponding to each driving speed are extracted respectively.
[0053] A coordinate system is constructed with driving speed as the horizontal coordinate and stitching ghost parameters, stitching misalignment parameters, loss image parameters and stitching gap parameters as the vertical coordinates. Several points are marked in the constructed coordinate system according to the arrangement results of driving speeds and the stitching ghost parameters, stitching misalignment parameters, loss image parameters and stitching gap parameters of the panoramic stitching image corresponding to each driving speed, to form a correlation curve between driving speed and stitching ghost, stitching misalignment, stitching loss and stitching gap.
[0054] In the further implementation of the above scheme, the stability analysis of the splicing quality is as follows: several driving speeds are uniformly selected on the horizontal axis of the correlation curve, and the splicing ghosting, splicing misalignment, splicing loss, and splicing gap change rate corresponding to each driving speed are extracted from the correlation curve in turn.
[0055] The absolute maximum values of splicing ghosting, splicing misalignment, splicing loss, and splicing gap change rates corresponding to each driving speed are extracted to obtain the absolute maximum splicing change rate and the absolute minimum splicing change rate corresponding to each driving speed. The absolute maximum splicing change rate is then subtracted from the absolute minimum splicing change rate to obtain the splicing change rate difference corresponding to each driving speed.
[0056] The absolute maximum splicing change rate corresponding to each driving speed is compared with adjacent driving speeds to calculate the splicing change fluctuation. The specific calculation is to subtract the absolute maximum splicing change rates of adjacent driving speeds, and calculate the standard deviation of all the obtained subtraction data as the splicing change fluctuation.
[0057] It should be added that the absolute maximum splicing change rate is selected from the splicing ghosting, splicing misalignment, splicing loss, and splicing gap change rates corresponding to each driving speed to calculate the splicing change fluctuation because the absolute maximum splicing change rate can capture the most serious changes, ensuring that the assessment of splicing quality fluctuations includes the possible worst case, thereby more directly and accurately reflecting the splicing change fluctuation.
[0058] The difference of splicing change rate corresponding to each driving speed is combined with the splicing change fluctuation to introduce into the analytical formula , get the splicing quality stability , where Indicates the splicing change fluctuation, Indicates The difference in splicing change rate corresponding to the driving speed, Indicates the driving speed number. , Represents a natural constant.
[0059] From the above analysis, it can be seen that the greater the difference in the splicing change rate corresponding to each driving speed, the greater the splicing change fluctuation, and the greater the stability of the splicing quality.
[0060] When analyzing the stitching quality stability of a 360-degree panoramic image at different driving speeds, the present invention obtains the stitching change rate difference at each driving speed by comparing the maximum values of stitching ghosting, stitching misalignment, stitching loss, and stitching gap change rate at each driving speed from the correlation curve of the stitching quality parameter and the driving speed, and performs stitching quality stability analysis in combination with the stitching change fluctuation calculated by the absolute maximum stitching change rate at adjacent driving speeds. Based on this, a more comprehensive and accurate analysis of the stitching quality stability of a panoramic image system can be provided from the changes between the stitching quality parameters at the same driving speed and the stitching changes between adjacent driving speeds.
[0061] S7. Determine the advantageous splicing quality parameter and the steady-state driving speed based on the formed correlation curve.
[0062] The process of determining the superior splicing quality parameters mentioned above is as follows: extract the absolute minimum values of splicing ghosting, splicing misalignment, splicing loss, and splicing gap change rate corresponding to each driving speed, and record the splicing quality parameters to which the absolute minimum splicing change rate corresponding to each driving speed belongs.
[0063] The splicing quality parameters corresponding to the absolute minimum splicing change rate of each driving speed are classified into the same splicing quality parameters, and the occurrence frequency of each splicing quality parameter is summarized, so as to extract the splicing quality parameter with the highest occurrence frequency as the dominant splicing quality parameter.
[0064] The present invention can identify the splicing quality parameters with the best performance under different speed conditions by extracting the absolute minimum splicing change rate under various driving speeds. Focusing on these optimal performance aspects can help optimize resource allocation and concentrate improvement work on the most advantageous parameters to improve optimization efficiency. At the same time, with the help of advantageous splicing quality parameters, the splicing quality parameters that are not advantageous can be identified, which in turn helps to formulate more targeted adjustment strategies.
[0065] Furthermore, the above-mentioned mid-steady-state driving speed determination process is as follows: inflection points are marked from the correlation curves of driving speed and splicing ghost, splicing misalignment, splicing loss, and splicing gap, and the marked inflection points divide the correlation curves into rising segments and non-rising segments, see Figure 3 shown.
[0066] It needs to be explained that the non-rising section includes the descending section and the horizontal section. The rising section of the correlation curve is that with the increase of driving speed, splicing ghosting, splicing misalignment, splicing loss and splicing gap will increase accordingly. The non-rising section of the correlation curve is that with the increase of driving speed, splicing ghosting, splicing misalignment, splicing loss and splicing gap will decrease while remaining stable. These are all optimistically accepted for splicing quality.
[0067] The driving speed intervals corresponding to the non-rising sections are respectively intercepted from the correlation curves between the driving speed and the splicing ghost, splicing misalignment, splicing loss and splicing gap.
[0068] An intersection operation is performed on the driving speed intervals corresponding to each non-ascending segment, and the operation result is used as the steady-state driving speed, wherein the steady-state driving speed obtained by the intersection operation can be a single value or an interval value.
[0069] The present invention extracts and determines the driving speed interval of the non-rising section from the correlation curves of various quality parameters, and then performs an intersection operation to obtain a steady-state driving speed. After determining the steady-state driving speed, the driving speed can be controlled to reduce quality defects such as splicing ghosting, misalignment, loss and gaps, thereby improving the overall splicing quality and reducing the risk of processing poor splicing. In addition, the steady-state driving speed obtained by comparing multiple curves and performing an intersection operation is a result based on data analysis, which makes the decision more scientific and reliable and reduces arbitrariness.
[0070] The driving speed and splicing ghost, splicing misalignment, splicing loss, and splicing gap correlation curves constructed by the present invention under the driving speed change scenario can not only be used to analyze the stability of splicing quality, but also be used to determine the dominant splicing quality parameters and the steady-state driving speed, so as to provide a decision based on reliable data for the optimization and adjustment of the splicing quality when the splicing quality is poor, reduce subjective judgment, improve the scientificity and accuracy of the decision, and have high practical value.
[0071] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.
Claims
1. A method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision, characterized in that: The system comprises a panoramic camera device, a speed sensor, an ambient lighting terminal and a processor, wherein the panoramic camera device is composed of cameras arranged in various directions, and the ambient lighting terminal is composed of multiple light sources, which evenly cover the field of view of the panoramic camera device, and specifically comprises the following steps: S1. Using the panoramic camera installed in the car to collect environmental images of each layout orientation in real time during the driving process of the car, and transmit them to the processor, which performs stitching processing on the transmitted environmental images of each layout orientation, and records the stitching completion time to form a panoramic stitching image time series; S2, using a speed sensor installed in the vehicle to detect the driving speed in real time during the driving process of the vehicle, and recording the driving speed detection time to form a driving speed time series; S3, mapping the driving speed time series with the panoramic stitching image time series, thereby establishing a corresponding relationship between the driving speed and the stitching duration; S4, respectively identifying the stitching ghost, stitching misalignment, loss image, and stitching gap of the panoramic stitching image time series, and obtaining the stitching quality parameters of each panoramic stitching image, specifically including the ghost parameter, the stitching misalignment parameter, the loss image parameter, and the stitching gap parameter; S5, comparing the driving speed with the splicing duration based on the driving speed time series and the corresponding relationship between the driving speed and the splicing duration, thereby analyzing the splicing processing stability; S6, forming a correlation curve between the stitching quality parameters of the panoramic stitching image time series and the driving speed time series, thereby analyzing the stitching quality stability; The correlation and comparison process between the driving speed and the splicing time is as follows: According to the driving speed time series, a number of points are marked in a coordinate system with time as the horizontal axis and driving speed as the vertical axis to form a driving speed change curve; In the formed driving speed variation curve, the splicing time is used as the ordinate, and the splicing time corresponding to each driving speed is marked according to the corresponding relationship between the driving speed and the splicing time to form a splicing time variation curve; The stability analysis of the splicing process is as follows: Mark the inflection points on the driving speed change curve and mark the driving speed corresponding to each inflection point; Based on the driving speed corresponding to each inflection point, the control points are marked on the splicing time change curve, and the tangent slope of each control point is obtained as the splicing time change rate of each control point; Based on the change rate of the splicing duration of each inflection point corresponding to the control point, the analytical formula Get the stability of splicing processing , where Indicates The change rate of the splicing duration of the inflection point corresponding to the control point, Indicates the inflection point number, ; The correlation curve is formed as follows: Arrange the driving speeds in the driving speed time series in order from small to large; According to the mapping group composed of the panoramic stitching image and the driving speed, the stitching ghost parameter, the stitching misalignment parameter, the loss image parameter and the stitching gap parameter of the panoramic stitching image corresponding to each driving speed are respectively extracted; A coordinate system is constructed with the driving speed as the horizontal coordinate and the splicing ghost parameter, the splicing misalignment parameter, the loss image parameter and the splicing gap parameter as the vertical coordinate. A number of points are marked in the constructed coordinate system according to the arrangement results of the driving speed and the splicing ghost parameter, the splicing misalignment parameter, the loss image parameter and the splicing gap parameter of the panoramic splicing image corresponding to each driving speed, so as to form a correlation curve between the driving speed and the splicing ghost, the splicing misalignment, the splicing loss and the splicing gap. The splicing quality stability analysis is as follows: A number of driving speeds are uniformly selected on the horizontal axis of the correlation curve, and the splicing ghost, splicing misalignment, splicing loss, and splicing gap change rate corresponding to each driving speed are sequentially extracted from the correlation curve; Extract the absolute maximum values of the splicing ghost, splicing dislocation, splicing loss, and splicing gap change rate corresponding to each driving speed to obtain the absolute maximum splicing change rate and the absolute minimum splicing change rate corresponding to each driving speed, and then subtract the absolute maximum splicing change rate from the absolute minimum splicing change rate to obtain the splicing change rate difference corresponding to each driving speed; The absolute maximum splicing change rate corresponding to each driving speed is compared with adjacent driving speeds to calculate the splicing change fluctuation; The difference of splicing change rate corresponding to each driving speed is combined with the splicing change fluctuation to introduce into the analytical formula , get the splicing quality stability , where Indicates the splicing change fluctuation, Indicates The difference in splicing change rate corresponding to the driving speed, Indicates the driving speed number. , Represents a natural constant.
2. The method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision as claimed in claim 1, characterized in that: The panoramic camera device and the speed sensor maintain a consistent acquisition frequency when acquiring environmental images and driving speed.
3. The method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision as claimed in claim 1, characterized in that: The process of establishing the corresponding relationship between the driving speed and the splicing time is as follows: The panoramic stitching images in the panoramic stitching image time series are numbered according to the arrangement order, and the driving speeds in the driving speed time series are numbered similarly; According to the one-to-one correspondence principle, the panoramic stitching images and the driving speeds with corresponding numbers are extracted from the panoramic stitching image time series and the driving speed time series to form a mapping group; Extracting the stitching completion time and driving speed detection time of the panoramic stitching images in each mapping group, thereby calculating the stitching duration corresponding to the driving speed in each mapping group; A corresponding relationship between the driving speed and the splicing duration is established based on the splicing duration corresponding to the driving speed in each mapping group.
4. The method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision as claimed in claim 1, characterized in that: The method also includes S7, determining a dominant splicing quality parameter and a steady-state driving speed based on the formed correlation curve.
5. The method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision as claimed in claim 4, characterized in that: The process of determining the superior splicing quality parameters is as follows: Extract the absolute minimum values of splicing ghost, splicing misalignment, splicing loss, and splicing gap change rate corresponding to each driving speed, and record the splicing quality parameters to which the absolute minimum splicing change rate corresponding to each driving speed belongs; The splicing quality parameters corresponding to the absolute minimum splicing change rate of each driving speed are classified into the same splicing quality parameters, and the occurrence frequency of each splicing quality parameter is summarized, so as to extract the splicing quality parameter with the highest occurrence frequency as the dominant splicing quality parameter.
6. The method for analyzing the quality of automobile 360-degree panoramic image stitching based on machine vision as claimed in claim 4, characterized in that: The steady-state driving speed determination process is as follows: Inflection points are marked from the correlation curves of driving speed and splicing ghost, splicing misalignment, splicing loss and splicing gap respectively, and the marked inflection points divide the correlation curves into rising sections and non-rising sections; The driving speed interval corresponding to the non-rising section is respectively intercepted from the correlation curves between the driving speed and the splicing ghost, splicing dislocation, splicing loss, and splicing gap; An intersection operation is performed on the driving speed intervals corresponding to each non-ascending segment, and the operation result is used as the steady-state driving speed.
