Machine vision detection system and method for road safety of environmental sanitation operation vehicle

Abstract

The invention discloses a sanitation operation vehicle road safety machine vision detection system and method, and the system comprises a vision detection system body which comprises a road detection platform, a vehicle self-inspection platform, a vehicle management platform, and a vision management platform. The road detection platform performs real-time detection on the road, the environment and the pedestrian behavior at the working position of the environmental sanitation working vehicle in real time; the vehicle self-inspection platform detects the condition of the sanitation operation vehicle before driving in real time; according to the invention, images around the vehicle body are seamlessly spliced into a bird's-eye view by setting the visual management platform, so that a real-time omnibearing visual angle is provided for a driver of the sanitation operation vehicle, and the phenomena of incomplete cleaning and even safety accidents caused by visual blind areas are avoided; the road detection platform is arranged to perform real-time detection and early warning on pedestrians, vehicles, road shapes and traffic conditions, so that threats caused by visual blind areas are reduced, and a powerful guarantee is provided for safe driving of environmental sanitation vehicles.

Description

Technical Field

[0001] The present invention relates to the technical field of road safety for sanitation operation vehicles, and in particular to a machine vision detection system and method for road safety for sanitation operation vehicles. Background Art

[0002] Sanitation vehicles are special vehicles with long bodies, large wheelbases, and high driving positions. There are large blind spots around the bodies of sanitation vehicles. In addition, there are differences between the inner and outer wheels, which make them special operation vehicles.

[0003] However, traditional sanitation vehicles have the following disadvantages when operating:

[0004] When driving a sanitation vehicle, due to the driver's limited visual range, blind spots can easily occur, resulting in incomplete cleaning and even safety accidents. Summary of the invention

[0005] The purpose of the present invention is to provide a machine vision detection system and method for road safety of sanitation vehicles, so as to solve the problem raised in the above background technology that in the process of driving sanitation vehicles, due to the limited visual range of the driver, blind spots may easily occur, resulting in incomplete cleaning and even safety accidents.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solutions: a machine vision inspection system for road safety of sanitation operation vehicles, comprising a visual inspection system body, wherein the visual inspection system body comprises a road inspection platform, a vehicle self-inspection platform, a vehicle management platform and a visual management platform;

[0007] The road detection platform performs real-time detection of the road, environment and pedestrian behavior at the location where the sanitation vehicle is operating;

[0008] The vehicle self-checking platform performs real-time detection of the condition of the sanitation operation vehicle before driving;

[0009] The vehicle management platform collects data, links equipment and displays real-time environment for sanitation vehicles;

[0010] The visual management platform performs visual sensing and detection of the environment of the sanitation vehicle from inside and outside the vehicle.

[0011] As a preferred technical solution of the present invention, the road detection platform includes a traffic sign line detection module, a traffic condition detection module, an illegal behavior detection module, a pedestrian behavior detection module and an obstacle detection module. The traffic sign line detection module performs position detection on various traffic sign lines, such as lanes, driving lanes, and parking lines. The traffic condition detection module performs real-time detection of traffic conditions and accurately measures the speed and number of vehicles to provide data support for traffic planning and road condition analysis. The illegal behavior detection module automatically identifies and records illegal parking behaviors on the road. The pedestrian behavior detection module performs real-time detection and prediction of pedestrian behavior on the road. The obstacle detection module automatically identifies various road markings and obstacles on the road surface.

[0012] As a preferred technical solution of the present invention, the vehicle self-inspection platform includes an oil level detection module, an attendance punching module, a GPS positioning module, a temperature and humidity module and a trajectory playback module. The oil level detection module performs real-time detection of the oil level in the sanitation vehicle, the attendance punching module performs real-time attendance punching for the sanitation vehicle driver, the GPS positioning module locates the position of the sanitation vehicle, the temperature and humidity module senses and detects the temperature and humidity inside and outside the sanitation vehicle in real time, and the trajectory playback module plays back the trajectory of the previous operation content of the sanitation vehicle.

[0013] As a preferred technical solution of the present invention, the vehicle management platform includes a data acquisition module, an over-limit alarm module, an equipment linkage module, a permission management module, a large-screen display module and a data storage module. The data acquisition module collects data of sanitation vehicles in real time, the over-limit alarm module warns of overload, the equipment linkage module links the sanitation vehicles with other sanitation equipment, the permission management module manages the permission to drive the sanitation vehicles, the large-screen display module displays the surrounding environment of the sanitation vehicles on a large screen, and the data storage module stores the managed data in real time.

[0014] As a preferred technical solution of the present invention, the visual management platform includes an outside-vehicle visual management module and an inside-vehicle visual management module. The outside-vehicle visual management module performs visual identification and management of the environment outside the sanitation vehicle, and the inside-vehicle visual management module performs visual identification and management of the environment inside the sanitation vehicle.

[0015] As a preferred technical solution of the present invention, the off-vehicle vision management module includes a feature recognition unit, a model establishment unit and a visual detection unit. The feature recognition unit performs feature recognition on images collected by off-vehicle vision, the model establishment unit establishes a road condition model of the off-vehicle environment, and the visual detection unit is a visual sensor installed on the outside of the sanitation vehicle.

[0016] As a preferred technical solution of the present invention, the feature recognition unit includes an image preprocessing component and a vehicle verification component. The image preprocessing component collects the external environment of the sanitation operation vehicle through a vision sensor, and is affected by lighting factors, reducing its imaging quality. Therefore, it is necessary to preprocess the region of interest. After preprocessing the image, useful information is highlighted and the interference of the background environment is removed. The preprocessing process includes image grayscale conversion, image grayscale enhancement, and filtering denoising. Since the acquired image is a color image with a large amount of information, in order to reduce the computational amount, it is necessary to first perform grayscale conversion on the original region of interest to obtain a grayscale image. After obtaining the grayscale image, the method of histogram equalization is used for grayscale enhancement to increase the global contrast of the image. At the same time, the grayscale-processed image often has noise interference, so it is necessary to filter the image. Since median filtering preserves image details to a certain extent and has a simple algorithm, median filtering is used to process the image. The calculation of the median filtering is as follows: When filtering a digital signal sequence xj (-∞ < j < ∞), first define a window with an odd length L, L = 2N + 1, N is a positive integer. At a certain moment, the signal samples within the window are x(i - N),..., x(i),..., x(i + N), where x(i) is the signal sample value located at the center of the window. After arranging these L signal sample values in ascending order, the sample value at the median at i is defined as the output value of the median filtering. The vehicle verification component sets R(x) as a one-dimensional function of the grayscale data of a certain row within the ROI region, so it is expressed in the form of an odd function and an even function. The symmetry measure of the corresponding region of interest is determined by the proportion of the even function separated from it to ensure the accuracy of feature extraction. The visual detection unit is provided with a vehicle body visual detector, a vehicle type recognition detector, a gap step detector, and a multi-station camera. The vehicle body visual detector detects the vehicle body of the sanitation operation vehicle, the vehicle type recognition detector detects the vehicle type of the sanitation operation vehicle, the gap step detector detects whether there is a gap on the surface of the sanitation vehicle and the width of the gap, and the multi-station camera is installed outside the sanitation operation vehicle to record the real-time external environment.

[0017] As a preferred technical solution of the present invention, the in-vehicle visual management module includes a positioning recognition unit, a preprocessing unit, and a segmentation unit. The positioning recognition unit automatically locates and recognizes, and obtains the coordinate data of the in-vehicle components in real time and updates the control system of the device in real time, so that the device completes the entire detection process according to the preset program. The preprocessing unit magnifies, filters, and performs edge detection on the collected image to extract the feature information in the image of the sanitation vehicle to meet subsequent recognition and processing. The segmentation unit processes the collected original image through a certain method to remove noise, performs region extraction and analysis on the image, and then judges the specific position of the workpiece to obtain the detection result.

[0018] The present invention provides a method for machine vision detection of road safety of sanitation operation vehicles, comprising the following steps:

[0019] Step 1: Image acquisition: The visual sensor installed inside the sanitation vehicle and the visual sensor installed outside the sanitation vehicle respectively collect the environment inside and outside the vehicle in real time;

[0020] Step 2: Model establishment: Establish corresponding SVM model and ANN model respectively;

[0021] Step 3: Image processing: The collected image is combined with the model to perform Gaussian blur, grayscale processing, Sobel operator (edge ​​detection), opening operation, closing operation, affine transformation, Holm linear detection and angle correction processing. The calculation formula of Holm linear detection is: the formula of a straight line passing through a point in the image is y0=x0*m+b, and this formula is a straight line in the parameter space. The calculation formula of angle correction processing is is the deflection angle;

[0022] Step 4: Verification and analysis: Verify the collected image again through character recognition and perform information analysis and processing on it;

[0023] Step 5. Detection and warning: The vehicle management platform issues real-time warnings to sanitation vehicles based on image information.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. By setting up a visual management platform, the images around the vehicle body are seamlessly spliced ​​into a bird's-eye view, providing a real-time all-round perspective for sanitation vehicle drivers, avoiding blind spots that cause incomplete cleaning and even safety accidents;

[0026] 2. By setting up a road detection platform, real-time detection and early warning of pedestrians, vehicles, road conditions and traffic conditions can be carried out respectively, reducing the threats caused by visual blind spots and becoming a strong guarantee for the safe driving of sanitation vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the architecture of the visual inspection system of the present invention;

[0028] Figure 2 It is a schematic diagram of the architecture of the road detection platform of the present invention;

[0029] Figure 3 It is a schematic diagram of the architecture of the vehicle self-checking platform of the present invention;

[0030] Figure 4 This is a schematic diagram of the architecture of the vehicle management platform of the present invention;

[0031] Figure 5 It is a schematic diagram of the architecture of the visual management platform of the present invention;

[0032] Figure 6 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with 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 making creative work are within the scope of protection of the present invention.

[0034] See also Figure 1-6 , the present invention provides a machine vision inspection system for road safety of sanitation operation vehicles, including a visual inspection system body, the visual inspection system body includes a road inspection platform, a vehicle self-inspection platform, a vehicle management platform and a visual management platform;

[0035] The road detection platform conducts real-time detection of the roads, environment and pedestrian behavior where sanitation vehicles are operating;

[0036] The vehicle self-inspection platform performs real-time inspection of the condition of sanitation vehicles before driving;

[0037] The vehicle management platform collects data, links equipment and displays real-time environment for sanitation vehicles;

[0038] The visual management platform performs visual sensing and detection of the environment of sanitation vehicles from inside and outside the vehicle.

[0039] The road detection platform includes a traffic sign detection module, a traffic condition detection module, an illegal behavior detection module, a pedestrian behavior detection module and an obstacle detection module. The traffic sign detection module detects the positions of various traffic sign markings, such as lanes, driving lanes and parking lines. The traffic condition detection module detects the traffic conditions in real time and accurately measures the speed and number of vehicles, providing data support for traffic planning and road condition analysis. The illegal behavior detection module automatically identifies and records illegal parking behaviors on the road. The pedestrian behavior detection module detects and predicts pedestrian behavior on the road in real time. The obstacle detection module automatically identifies various road markings and obstacles on the road surface.

[0040] The vehicle self-inspection platform includes an oil level detection module, an attendance punching module, a GPS positioning module, a temperature and humidity module, and a trajectory playback module. The oil level detection module performs real-time detection of the oil level in the sanitation vehicle. The attendance punching module performs real-time attendance punching for the sanitation vehicle drivers. The GPS positioning module locates the position of the sanitation vehicle. The temperature and humidity module senses and detects the temperature and humidity inside and outside the sanitation vehicle in real time. The trajectory playback module plays back the trajectory of the previous operation content of the sanitation vehicle.

[0041] The vehicle management platform includes a data acquisition module, an over-limit alarm module, an equipment linkage module, a permission management module, a large-screen display module and a data storage module. The data acquisition module collects data on sanitation vehicles in real time, the over-limit alarm module issues an early warning of overload, the equipment linkage module links sanitation vehicles with other sanitation equipment, the permission management module manages the permission to drive sanitation vehicles, the large-screen display module displays the surrounding environment of sanitation vehicles on a large screen, and the data storage module stores the managed data in real time.

[0042] The visual management platform includes an external visual management module and an internal visual management module. The external visual management module performs visual identification and management of the environment outside the sanitation vehicle, and the internal visual management module performs visual identification and management of the environment inside the sanitation vehicle.

[0043] The off-vehicle vision management module includes a feature recognition unit, a model establishment unit and a visual detection unit. The feature recognition unit performs feature recognition on the images collected by the off-vehicle vision, the model establishment unit establishes a road condition model of the off-vehicle environment, and the visual detection unit is a visual sensor installed on the outside of the sanitation vehicle.

[0044] The feature recognition unit includes an image preprocessing component and a vehicle verification component. The image preprocessing component collects the external environment of the sanitation operation vehicle through a vision sensor, and is affected by lighting factors, reducing its imaging quality. Therefore, it is necessary to preprocess the region of interest. Through the preprocessed image, useful information is highlighted and the interference of the background environment is removed. The preprocessing process includes image grayscale conversion, image grayscale enhancement, and filtering denoising. Since the acquired image is a color image with a large amount of information, in order to reduce the computational amount, it is necessary to first perform grayscale conversion on the original region of interest to obtain a grayscale image. After obtaining the grayscale image, the method of histogram equalization is used for grayscale enhancement to increase the global contrast of the image. At the same time, the grayscale-processed image often has noise interference, so it is necessary to filter the image. Since median filtering preserves image details to a certain extent and the algorithm is simple, median filtering is used to process the image. The calculation of median filtering is as follows: When filtering a digital signal sequence xj (-∞ < j < ∞), first define a window of length L, where L is an odd number and L = 2N + 1, N is a positive integer. At a certain moment, the signal samples within the window are x(i - N),..., x(i),..., x(i + N), where x(i) is the signal sample value located at the center of the window. After arranging these L signal sample values in ascending order, the sample value at the median at i is defined as the output value of the median filtering. The vehicle verification component sets R(x) as a one-dimensional function of the grayscale data of a certain row within the ROI region, so it is expressed in the form of an odd function and an even function. The symmetry measure of the corresponding region of interest is determined by the proportion of the even function separated from it to ensure the accuracy of feature extraction. The vision detection unit is equipped with a vehicle body vision detector, a vehicle type recognition detector, a gap and step detector, and a multi-station camera. The vehicle body vision detector detects the vehicle body of the sanitation operation vehicle, the vehicle type recognition detector detects the vehicle type of the sanitation operation vehicle, the gap and step detector detects whether there are gaps on the surface of the sanitation vehicle and the width of the gaps, and the multi-station camera is installed outside the sanitation operation vehicle to record the real-time external environment.

[0045] The in-vehicle vision management module includes a positioning and recognition unit, a preprocessing unit, and a segmentation unit. The positioning and recognition unit automatically locates and recognizes, and obtains the coordinate data of the in-vehicle components in real time and updates the control system of the device in real time, so that the device completes the entire detection process according to the preset program. The preprocessing unit performs magnification, filtering, and edge detection processing on the collected images to extract the feature information in the images of the sanitation vehicle to meet subsequent recognition and processing. The segmentation unit processes the collected original images through a certain method, removes noise, performs region extraction and analysis on the images, and then judges the specific position of the workpiece to obtain the detection result.

[0046] A method for machine vision detection of road safety of a sanitation operation vehicle according to the present invention includes the following steps:

[0047] Step 1: Image acquisition: The visual sensor installed inside the sanitation vehicle and the visual sensor installed outside the sanitation vehicle respectively collect the environment inside and outside the vehicle in real time;

[0048] Step 2: Model establishment: Establish corresponding SVM model and ANN model respectively;

[0049] Step 3: Image processing: The collected image is combined with the model to perform Gaussian blur, grayscale processing, Sobel operator (edge ​​detection), opening operation, closing operation, affine transformation, Holm linear detection and angle correction processing. The calculation formula of Holm linear detection is: the formula of a straight line passing through a point in the image is y0=x0*m+b, and this formula is a straight line in the parameter space. The calculation formula of angle correction processing is is the deflection angle;

[0050] Step 4: Verification and analysis: Verify the collected image again through character recognition and perform information analysis and processing on it;

[0051] Step 5. Detection and warning: The vehicle management platform issues real-time warnings to sanitation vehicles based on image information.

[0052] In the present invention, the visual sensor installed in the sanitation vehicle and the visual sensor installed outside the sanitation vehicle respectively collect the environment inside and outside the vehicle in real time; the corresponding SVM model and ANN model are respectively established; the collected images are combined with the model to perform Gaussian blur, grayscale processing, Sobel operator (edge ​​detection), opening operation, closing operation, affine transformation, Holm linear detection and angle correction processing, and the Holm linear detection calculation formula is: the formula of the straight line passing through a certain point in the image is y0=x0*m+b, and this formula is a straight line in the parameter space, and the calculation formula for the angle correction processing is is the deflection angle; the collected image is verified twice through character recognition and information analysis and processing is performed on it; the vehicle management platform issues real-time operation warnings to sanitation vehicles based on the image information.

[0053] In the present invention, the traffic sign line detection module detects the positions of various traffic sign lines, such as lanes, driving lanes, and parking lines; the traffic condition detection module detects the traffic conditions in real time and accurately measures the speed and number of vehicles, providing data support for traffic planning and road condition analysis; the illegal behavior detection module automatically identifies and records illegal parking behaviors on the road; the pedestrian behavior detection module detects and predicts pedestrian behavior on the road in real time; the obstacle detection module automatically identifies various road markings and obstacles on the road surface; the oil level detection module detects the oil level in the sanitation vehicle in real time; the attendance punching module punches in and out the drivers of the sanitation vehicles in real time; the GPS positioning module locates the position of the sanitation vehicle; and the temperature and humidity The module senses and detects the temperature and humidity inside and outside the sanitation vehicle in real time. The trajectory playback module plays back the previous operation content of the sanitation vehicle. The data acquisition module collects data of the sanitation vehicle in real time. The over-limit alarm module warns of overload. The equipment linkage module connects the sanitation vehicle with other sanitation equipment. The authority management module manages the authority to drive the sanitation vehicle. The large-screen display module displays the surrounding environment of the sanitation vehicle on a large screen. The data storage module stores the managed data in real time. The outside-vehicle visual management module visually identifies and manages the environment outside the sanitation vehicle. The inside-vehicle visual management module visually identifies and manages the environment inside the sanitation vehicle.

[0054] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A machine vision inspection system for road safety of sanitation vehicles, comprising a visual inspection system body, characterized in that: The visual inspection system body includes a road inspection platform, a vehicle self-inspection platform, a vehicle management platform and a visual management platform; The road detection platform performs real-time detection of the road, environment and pedestrian behavior at the location where the sanitation vehicle is operating; The vehicle self-checking platform performs real-time detection of the condition of the sanitation operation vehicle before driving; The vehicle management platform collects data, links equipment and displays real-time environment for sanitation vehicles; The visual management platform performs visual sensing and detection of the environment of the sanitation vehicle from inside and outside the vehicle.

2. A road safety machine vision detection system for sanitation vehicles according to claim 1, characterized in that: The road detection platform includes a traffic sign line detection module, a traffic condition detection module, an illegal behavior detection module, a pedestrian behavior detection module and an obstacle detection module. The traffic sign line detection module performs position detection on various traffic sign lines, such as lanes, driving lanes and parking lines. The traffic condition detection module performs real-time detection of traffic conditions and accurately measures the speed and number of vehicles to provide data support for traffic planning and road condition analysis. The illegal behavior detection module automatically identifies and records illegal parking behaviors on the road. The pedestrian behavior detection module performs real-time detection and prediction of pedestrian behavior on the road. The obstacle detection module automatically identifies various road markings and obstacles on the road surface.

3. A road safety machine vision detection system for sanitation vehicles according to claim 1, characterized in that: The vehicle self-inspection platform includes an oil level detection module, an attendance punching module, a GPS positioning module, a temperature and humidity module and a trajectory playback module. The oil level detection module performs real-time detection of the oil level in the sanitation vehicle, the attendance punching module performs real-time attendance punching for the sanitation vehicle driver, the GPS positioning module locates the position of the sanitation vehicle, the temperature and humidity module senses and detects the temperature and humidity inside and outside the sanitation vehicle in real time, and the trajectory playback module plays back the trajectory of the previous operation content of the sanitation vehicle.

4. A road safety machine vision detection system for sanitation vehicles according to claim 1, characterized in that: The vehicle management platform includes a data acquisition module, an over-limit alarm module, an equipment linkage module, a permission management module, a large-screen display module and a data storage module. The data acquisition module collects data of sanitation vehicles in real time, the over-limit alarm module warns of overload, the equipment linkage module links sanitation vehicles with other sanitation equipment, the permission management module manages the permission to drive sanitation vehicles, the large-screen display module displays the surrounding environment of sanitation vehicles on a large screen, and the data storage module stores the managed data in real time.

5. The machine vision detection system for road safety of sanitation vehicles according to claim 1, characterized in that: The visual management platform includes an off-vehicle visual management module and an on-vehicle visual management module. The off-vehicle visual management module performs visual recognition and management on the environment outside the sanitation vehicle, and the on-vehicle visual management module performs visual recognition and management on the environment inside the sanitation vehicle.

6. A road safety machine vision detection system for sanitation vehicles according to claim 5, characterized in that: The off-vehicle visual management module includes a feature recognition unit, a model establishment unit and a visual detection unit. The feature recognition unit performs feature recognition on images collected from the off-vehicle visual environment. The model establishment unit establishes a road condition model of the off-vehicle environment. The visual detection unit is a visual sensor installed on the outside of the sanitation vehicle.

7. A road safety machine vision detection system for sanitation vehicles according to claim 6, characterized in that: The feature recognition unit includes an image preprocessing component and a vehicle verification component. The image preprocessing component acquires the external environment of the sanitation operation vehicle through a vision sensor. However, it is affected by lighting factors, reducing its imaging quality. Therefore, it is necessary to preprocess the region of interest. After preprocessing the image, useful information is highlighted, and the interference of the background environment is removed. The preprocessing process includes image grayscale conversion, image grayscale enhancement, and filtering denoising. Since the acquired image is a color image with a large amount of information, in order to reduce the computational amount, it is necessary to first perform grayscale conversion on the original region of interest to obtain a grayscale image. After obtaining the grayscale image, the method of histogram equalization is used for grayscale enhancement to increase the global contrast of the image. At the same time, the image after grayscale processing often has noise interference. Therefore, it is necessary to filter the image. Since median filtering preserves image details to a certain extent and has a simple algorithm, median filtering is adopted to process the image. The calculation of the median filtering is as follows: When filtering a digital signal sequence xj (-∞ < j < ∞), first, a window with an odd length L is defined, L = 2N + 1, where N is a positive integer. At a certain moment, the signal samples within the window are x(i - N), …, x(i), …, x(i + N), where x(i) is the signal sample value located at the center of the window. After arranging these L signal sample values in ascending order, the sample value at the median at i is defined as the output value of the median filtering. The vehicle verification component sets R(x) as a one-dimensional function of the grayscale data of a certain row within the ROI region. Therefore, it is expressed in the form of an odd function and an even function. The symmetry measure of the corresponding region of interest is determined by the proportion of the even function separated from it, ensuring the accuracy of feature extraction. The visual detection unit is provided with a vehicle body visual detector, a vehicle type recognition detector, a gap step detector, and a multi-station camera. The vehicle body visual detector detects the vehicle body of the sanitation operation vehicle. The vehicle type recognition detector detects the vehicle type of the sanitation operation vehicle. The gap step detector detects whether there are gaps on the surface of the sanitation vehicle and the width of the gaps. The multi-station camera is installed outside the sanitation operation vehicle to record the real-time external environment.

8. The machine vision detection system for road safety of sanitation vehicles according to claim 5, characterized in that: The in-vehicle visual management module includes a positioning and recognition unit, a preprocessing unit, and a segmentation unit. The positioning and recognition unit automatically locates and recognizes, and real-time obtains the coordinate data of the in-vehicle components and updates the control system of the device in real time, enabling the device to complete the entire detection process according to the preset program. The preprocessing unit performs magnification, filtering, and edge detection processing on the collected image to extract the feature information in the image of the sanitation vehicle to meet subsequent recognition and processing. The segmentation unit processes the collected original image through a certain method to remove noise, performs region extraction and analysis on the image, and then judges the specific position of the workpiece to obtain the detection result.

9. A method for detecting road safety of sanitation vehicles by machine vision according to any one of claims 1 to 8, characterized in that: It includes the following steps: Step 1, image acquisition: The vision sensors installed inside and outside the sanitation operation vehicle respectively collect the internal and external environments of the vehicle in real time; Step 2: Model establishment: Establish corresponding SVM model and ANN model respectively; Step 3: Image processing: The collected image is combined with the model to perform Gaussian blur, grayscale processing, Sobel operator (edge ​​detection), opening operation, closing operation, affine transformation, Holm linear detection and angle correction processing. The calculation formula of Holm linear detection is: the formula of a straight line passing through a point in the image is y0=x0*m+b, and this formula is a straight line in the parameter space. The calculation formula of angle correction processing is is the deflection angle; Step 4: Verification and analysis: Verify the collected image again through character recognition and perform information analysis and processing on it; Step 5. Detection and warning: The vehicle management platform issues real-time warnings to sanitation vehicles based on image information.

Images