A method and system for ultra-fast analysis of mixed waste plastic components

Abstract

The application discloses a kind of mixed waste plastic component ultrafast analysis method and system, it is related to environmental protection technical field, including obtaining the image information of target area and carrying out pretreatment;YOLOv5 algorithm training model is used, obtains waste fragment position information and divides picture;The shortest scanning path based on the position information is planned, and the composition of the fragment is analyzed using the near-infrared spectrum obtained by scanning;Based on the segmented picture, the corresponding optimal threshold is predicted using a convolutional neural network model, and edge detection is performed using a Canny operator;The edge detection result is subjected to contour extraction and area calculation to obtain the area corresponding to the fragment.The application realizes the rapid positioning and detection of multiple waste plastic fragments on the same plane, improving the detection speed of mixed waste plastic;Using edge detection, infrared spectrum analysis and other technologies, the composition and proportion of mixed waste plastic are analyzed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of environmental protection, and in particular to a mixed waste plastic component ultra-fast analysis method and system. BACKGROUND

[0002] Due to the stability of the plastic structure compared to general waste, the average natural degradation time is 300 to 500 years. In recent years, with the development of industry and the improvement of people's living standards, the annual output of global plastic garbage has gradually increased, and the plastic garbage that has not been effectively treated will cause great harm to the earth's environment. Therefore, it is necessary to recycle and regenerate plastic garbage. However, the waste plastic obtained by recycling is often mixed, and direct treatment may cause secondary environmental hazards; and the regenerated plastic obtained by directly using mixed waste plastic for granulation also has poor performance, and even contains harmful ingredients. Therefore, it is necessary to analyze the components of mixed waste plastic.

[0003] At present, there is a portable mixed plastic component rapid analysis device, which includes a foldable analysis device top cover and a bottom cover, the top cover is filled with EVA foam inside, used for placing a scanning device, an X-axis moving device, an external cable and a matching tool; the bottom cover seals a terminal computer, a power supply and a Y-axis moving device in its interior by a sealing plate, and the X-axis moving device and the scanning device are installed above the Y-axis moving device during work. When the work is completed, the X-axis moving device and the scanning device are detached from the fixed position on the bottom cover of the analysis device and placed in the top cover of the analysis device, fixed by the EVA foam, and the top cover and the bottom cover of the analysis device are folded and closed, locked as a suitcase state. The device integrates spectrum recognition and automatic control, has high automation degree, large processing capacity and is suitable for industrial production. However, the device needs to scan each point on the whole plane every time, which is time-consuming and inefficient, and cannot realize ultra-fast analysis of mixed waste plastic components. SUMMARY

[0004] In view of the problems in the existing mixed waste plastic component ultra-fast analysis method, the present application is proposed.

[0005] Therefore, the problem to be solved by the present application is that the existing device needs to scan each point on the whole plane every time, which is time-consuming and inefficient, and cannot realize ultra-fast analysis of mixed waste plastic components.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the embodiments of the present application provide a mixed waste plastic component ultra-fast analysis method, which comprises: obtaining image information of a target area and preprocessing the image information; performing anchor frame on the preprocessed image information to obtain waste fragment position information and segment the picture; performing optimal path planning; moving based on the optimal path using an infrared scanning module, performing infrared scanning, and analyzing the composition of the waste fragments according to spectral information; using a convolutional neural network model to predict the optimal Canny edge detection operator threshold value, and performing edge detection and contour extraction to calculate the area of the waste fragments; and analyzing the mixed waste plastic components based on the waste fragment composition information and the waste fragment area information.

[0008] As a preferred scheme of the mixed waste plastic component ultra-fast analysis method, the preprocessing comprises: obtaining an RGB image of a mixed waste plastic sample, and performing picture cropping to make the RGB image correspond to the range of the target area.

[0009] As a preferred scheme of the mixed waste plastic component ultra-fast analysis method, the obtaining of the waste fragment position information comprises: establishing a data set composed of RGB images of mixed waste samples, and labeling waste plastic fragments in the images; training a YOLOv5 model using the labeled data set; and performing anchor frame on the RGB images based on the trained YOLOv5 model to determine the waste fragment position information.

[0010] As a preferred scheme of the mixed waste plastic component ultra-fast analysis method, the planning of the optimal path comprises: combining a dynamic programming algorithm and a greedy algorithm to calculate a path with the shortest time for traversing all positions as the optimal path; controlling the infrared scanning module to move along the optimal path and stop at the waste fragments to perform scanning; and using a classification model to further obtain fragment composition information at the point based on the near-infrared spectrum obtained by scanning.

[0011] As a preferred scheme of the mixed waste plastic component ultra-fast analysis method, the prediction of the threshold value comprises: training a convolutional neural network model to predict the optimal threshold value of a Canny edge detection operator corresponding to a picture; predicting the optimal threshold value of a Canny edge detection operator corresponding to the segmented picture based on the segmented picture using the convolutional neural network model; performing edge detection on the segmented picture using the Canny operator and the predicted optimal threshold value; performing inflation and corrosion operations on the detected edge information to obtain complete and closed contour lines of each fragment; and performing contour area calculation based on the closed contour to obtain fragment area information.

[0012] As a preferred scheme of the mixed waste plastic component ultrafast analysis method, wherein: the image information is subjected to convolution operation based on the Canny operator to obtain gradient information; expressions of the Canny operator in x and y directions are as follows respectively:

[0013]

[0014]

[0015] Wherein, S x For calculating the image x direction pixel gradient matrix G x ,S y For calculating the image y direction pixel gradient matrix G y , I is a gray image matrix, and * represents cross-correlation calculation; a gradient strength matrix G xy :

[0016]

[0017] Wherein, g x (i,j) is an element located at the i-th row and the i-th column in the G x matrix, g y (i,j) is an element located at the i-th row and the i-th column in the G y matrix, g xy (i,j) is an element located at the i-th row and the j-th column in the G xy matrix.

[0018] As a preferred scheme of the mixed waste plastic component ultrafast analysis method, wherein: the analysis of the waste mixed plastic component is based on the fragment component information and the area information, the proportion of each component in the mixed waste plastic is analyzed, the original point-by-point scanning scheme is replaced by the optimal path planning, the time required for one scanning is greatly reduced, and thus the ultrafast component analysis is realized.

[0019] In a second aspect, the embodiment of the present application provides a mixed waste plastic component ultrafast analysis system, which comprises: a collection module, which acquires image information of a target area and pre-processes the image information, and obtains waste fragment position information and divides a picture by performing anchor frame on the pre-processed image information and performing optimal path planning; a scanning module, which moves based on the optimal path and performs infrared scanning by using an infrared scanning module, and analyzes components of the waste fragments according to spectral information; a calculation module, which uses a convolution neural network model to predict a best Canny edge detection operator threshold value, and performs edge detection and contour extraction, and calculates areas of the waste fragments; and an analysis module, which analyzes waste mixed plastic components based on the waste fragment component information and the waste fragment area information.

[0020] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory and a processor, the memory storing a computer program, wherein the processor executes the computer program to implement any step of the mixed waste plastic composition ultrafast analysis method described above.

[0021] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, storing a computer program, wherein the computer program is executed by a processor to implement any step of the mixed waste plastic composition ultrafast analysis method described above.

[0022] The present application has the advantages that the present application uses image segmentation and path planning technology to realize rapid positioning and detection of multiple waste plastic fragments on the same plane, improves the detection speed of mixed waste plastic, and saves time. The present application uses edge detection, infrared spectrum analysis and other technologies to realize mixed waste plastic composition type and proportion analysis. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Among them:

[0024] Figure 1 It is a flowchart of the mixed waste plastic composition ultrafast analysis method;

[0025] Figure 2 It is an example diagram of the pretreated mixed waste plastic sample;

[0026] Figure 3 It is an example diagram of the sample image after segmentation. DETAILED DESCRIPTION

[0027] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0028] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the scope of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0029] Second, the "one embodiment" or "an embodiment" referred to herein means a specific feature, structure, characteristic, or combination thereof found in at least one implementation of the present application. The various appearances of "in one embodiment" or "in an embodiment" in the specification do not all refer to the same embodiment, although they can.

[0030] Embodiment 1

[0031] Reference Figures 1-3 For the first embodiment of the present application, the embodiment provides a mixed waste plastic component ultra-fast analysis method, comprising:

[0032] S1: Obtain image information of the target area and pre-process it, and anchor the pre-processed image information to obtain waste fragment position information and segment the picture, and perform optimal path planning.

[0033] Use the built-in camera of the mobile phone to take pictures of the target area and obtain the RGB image of the mixed waste plastic sample;

[0034] By cropping the picture, the excess background in the picture can be removed, so that the RGB image completely corresponds to the target area range, avoiding interference with subsequent processing.

[0035] Get a complete, clear and clean mixed waste plastic sample image as shown in Figure 2 .

[0036] Establish a data set composed of RGB images of mixed waste samples, and use labelimg software to label the waste plastic fragments in the image;

[0037] Use the labeled data set to train the YOLOv5 model;

[0038] Use the trained YOLOv5 model to anchor the RGB image taken by the mobile phone to determine the waste fragment position information.

[0039] Use labelimg software to label:

[0040] As shown in Figure 2 , a complete mixed waste plastic sample image contains multiple fragments, which may contain many different materials, including waste plastic fragments of PET, PP, PS, PVC, ABS, etc., and may also contain non-plastic impurities such as paper scraps and films. When using labelimg software to label, the waste plastic fragments are labeled as "plastic" class, and the impurities are not labeled, so that the model can learn the visual features of the waste plastic fragments.

[0041] Use the labeled data set to train the YOLOv5 model:

[0042] YOLOv5 is a single-stage object detection algorithm, and the network architecture of the object detection algorithm mainly includes four parts: input, baseline network, Neck network and Head output. The input represents the input picture, and the input image size of this network is 608*608. The input image needs to be scaled to the input size of the network first, and normalized and other operations. The baseline network is a network composed of some excellent classifiers. This module is used to extract some general feature representations. In YOLOv5, CSPDarknet53 structure and Focus structure are used as the baseline network. The Neck network is located in the middle position between the baseline network and the head network, which is used to further improve the diversity and robustness of the features. The Neck network of YOLOv5 adopts the FPN+PAN structure, and makes improvements on this basis to strengthen the network feature fusion capability. Head is used to complete the output of the target detection result. YOLOv5 uses GIOU_Loss loss function to further improve the detection accuracy of the algorithm. The model is trained using the labeled dataset, and finally the detection accuracy of all sample images of waste plastic fragments reaches 99.5%.

[0043] The trained YOLOv5 model is used to segment the picture and determine the position information of the waste fragment:

[0044] The trained YOLOv5 model is saved, and the model is used to detect the preprocessed mixed waste plastic sample image to obtain the picture segmentation result as shown in Figure 3 and the file containing the coordinate information of each waste plastic fragment in the image as shown in Table 1.

[0045] The optimal path planning needs to preprocess the coordinates of the waste plastic fragments in advance:

[0046] As shown in Table 1, the txt file output by the YOLOv5 model contains 6 columns of data, and each row corresponds to the relevant information of a detected waste plastic fragment.

[0047] Table 1 Waste plastic fragment coordinate information

[0048] Class Abscissa Ordinate Anchor box length Anchor box width Confidence 0 0.796549 0.879883 0.0507812 0.036458 0.827174

[0049] The output format is as follows:

[0050] The first column represents the class index of the detected object. In this embodiment, the detection target only has one class "plastic", so the class index column in the output file is all 0;

[0051] The two numbers in the second and third columns represent the relative position of the center point of the bounding box of the detected object in the image. The numbers here are normalized, that is, the coordinate values are divided by the width and height of the image, so they are all in the range of 0 to 1. The first number is the X coordinate (horizontal position) of the center point, and the second number is the Y coordinate (vertical position).

[0052] The two numbers in the fourth and fifth columns represent the width and height of the bounding box of the detected object, which are also normalized values relative to the image size.

[0053] The number in the last column represents the confidence of the model for this detection result, which usually indicates the degree of confidence that the model is confident that the detected object belongs to the predicted category. The closer the confidence value is to 1, the more confident the model is about the accuracy of its prediction.

[0054] Therefore, the data in the second and third columns need to be multiplied by the width and length of the picture to obtain coordinate values corresponding to the actual size of the image.

[0055] Secondly, regarding the path planning algorithm, it needs to be explained that:

[0056] The problem involved in this embodiment is to find the shortest path that traverses all the points contained in the coordinate information from a fixed starting point without returning to the starting point. If all possible paths are considered and exhausted, the complexity is n!, which is obviously too high in terms of calculation cost and too long in terms of time, and is not suitable for practical situations. Therefore, this embodiment combines a more accurate dynamic programming algorithm with a shorter approximate algorithm (greedy algorithm).

[0057] When the number of fragments in the picture is small, a dynamic programming algorithm is used to solve the path planning problem. The main purpose is to find an accurate shortest path in a given set of coordinates, so that each point is passed exactly once. The dynamic programming algorithm uses the optimal solution of the subproblem to construct the optimal solution of the original problem, thereby effectively reducing the computational complexity.

[0058] The dividing point between less and more is 20, and when the number of fragments is less than or equal to 20, the dynamic programming algorithm is used to calculate the optimal path, and when the number of fragments is greater than 20, the greedy algorithm is used to calculate the optimal path. This judgment standard is based on the minimum setting of the sum of the path calculation time and the actual moving time. The dynamic programming algorithm obtains the absolute shortest path, while the greedy algorithm obtains the approximate shortest path, which leads to the fact that the path obtained by the dynamic programming algorithm will save a certain actual moving time; at the same time, the calculation time of the dynamic programming algorithm increases exponentially with the increase of the number of fragments, while the calculation time of the greedy algorithm is stable. After many experiments, when the number of fragments is less than or equal to 20, the sum of the calculation path time and the actual moving time using the dynamic programming algorithm is usually smaller; when the number of fragments is greater than 20, the increased calculation time of the dynamic programming algorithm has exceeded the saved moving time, so the greedy algorithm is better.

[0059] The specific steps are as follows:

[0060] Define the subproblem: Let C(S, j) represent the shortest path length from the starting point to all coordinates in set S. Where S is a set of coordinates containing the starting point, and j is a coordinate in S.

[0061] Initialization: For the subproblem with only the starting point, it is obvious that C({0}, 0) = 0, and the initial value of other cases is infinity.

[0062] State transition equation: for all S and j, if j belongs to S, and S contains at least two coordinates, the state transition equation is:

[0063] C(S, j) = min{C(S-{j}, i) + d ij |i∈S,i≠j}

[0064] Where d ij represents the distance between coordinates i and j.

[0065] Calculate the optimal solution: by gradually increasing the size of the set S, the value of C(S, j) is calculated from small to large, and the optimal solution of the whole coordinate set is finally obtained.

[0066] Reconstruct the optimal path: the final result is obtained by tracing back the selection path of the optimal subproblem.

[0067] When the number of fragments in the picture is large, the greedy algorithm is used to solve the path planning problem, and the basic idea is to select the current optimal solution at each step, so as to hope to achieve global optimization through a series of local optimal selection. Although the greedy algorithm cannot guarantee to find the global optimal solution, its calculation speed is faster, and it is suitable for approximate solution of large-scale problems. The specific steps include:

[0068] Initial setting: start from a starting point, and the coordinate is marked as visited.

[0069] Greedy selection: at each step, starting from the current coordinate, the nearest unvisited coordinate to the current coordinate is selected as the next target for visiting, and the coordinate is marked as visited.

[0070] Repeat the above steps to continue selecting the nearest unvisited coordinate from the new current coordinate until all coordinates are visited.

[0071] S2: using an infrared scanning module to move based on the optimal path, performing infrared scanning, and analyzing the composition of the waste fragments based on spectral information.

[0072] Controlling the infrared scanning module to move along the optimal path, performing infrared scanning on the fragments, and analyzing the composition of the waste fragments based on spectral information. It should be noted that:

[0073] Currently, there is a device that can control the infrared scanning module to move along the optimal path and stop at the waste fragments for scanning through the motor;

[0074] The near-infrared spectral analysis includes using the existing classification model to further obtain the fragment composition information at the point based on the scanned near-infrared spectrum.

[0075] S3: using a convolutional neural network model to predict its optimal Canny edge detection operator threshold, and performing edge detection and contour extraction to calculate the area of the waste fragments.

[0076] Since the background color of the sample picture in the corresponding use scenario of the present embodiment is usually white, and most of the waste plastic fragments are also light-colored, the similar colors result in poor edge detection effect when directly using the Canny operator in the OpenCV library. Therefore, the present application adopts a deep learning method to train a model for predicting the optimal threshold corresponding to a picture. The model is modified based on the VGG16 model to obtain an optimal threshold prediction model, which inputs a sample picture and outputs the threshold corresponding to the picture that can make the Canny edge detection operator obtain the best result.

[0077] The key parameters of the Canny operator are two thresholds, which determine the edge detection effect. Since the goal of this step is to obtain the area of the fragments through edge detection and a series of image operations, the so-called "optimal threshold" is the threshold that gives the most accurate area after edge detection. In order to achieve this goal, first, 500 fragment images are obtained by shooting and image segmentation, then 81*201=16281 combinations of low threshold [0,80] and high threshold [40,240] are traversed, and the area obtained by edge detection and a series of image operations under each threshold setting is compared with the accurate area obtained by manually painting the fragments black, and the threshold with the smallest relative error and an absolute value not exceeding 0.1 is taken as the optimal threshold in the data set. After these preparations, a total of 379 data sets consisting of "picture-optimal threshold" are obtained. Using this data set and based on the VGG16 model, a regression model is created for training, and the optimal threshold prediction model is obtained. By inputting the picture, the threshold that gives the best result for the Canny edge detection operator can be obtained.

[0078] For the segmented image obtained in the above step, the corresponding Canny edge detection optimal threshold is predicted based on the model, and then edge detection is performed using the Canny operator.

[0079] Among them, the edge detection includes using the Canny operator and the optimal threshold to perform edge detection on the segmented image.

[0080] The area calculation includes performing dilation and erosion operations on the detected edge information to obtain the complete and closed contour line of each fragment.

[0081] Based on the closed contour, the OpenCV library is used to calculate the area of the contour to obtain the fragment area information.

[0082] The Canny operator is used to perform convolution operation on the image information to obtain gradient information.

[0083] The expressions of the Canny operator in x and y directions are as follows:

[0084]

[0085]

[0086] Where S x The Canny operator is used to perform convolution operation on the image information to obtain gradient information. x The Canny operator is used to perform convolution operation on the image information to obtain gradient information. y The Canny operator is used to perform convolution operation on the image information to obtain gradient information. y I is a grayscale image matrix, and * represents cross-correlation calculation.

[0087] The gradient intensity matrix Gxy is obtained by the following calculation.

[0088]

[0089] Where g x (i,j) is the element in the i-th row and j-th column of the G x matrix, g y (i,j) is the element in the i-th row and i-th column of the G y matrix, g xy (i,j) is the element in the i-th row and j-th column of the G xy matrix.

[0090] S4: analyzing the components of the waste mixed plastics based on the waste fragment component information and the waste fragment area information.

[0091] Based on the fragment component information and area information, the proportion of each component in the mixed waste plastics is analyzed.

[0092] The optimal path planning is used to replace the original point-by-point scanning scheme, so that the time required for one scanning is greatly reduced, and super-fast component analysis is realized.

[0093] In summary, the present application uses image segmentation, path planning and other technologies to realize the rapid positioning and detection of multiple waste plastic fragments on the same plane, improves the detection speed of mixed waste plastics, saves time, uses edge detection, infrared spectrum analysis and other technologies to realize the analysis of the component types and proportions of mixed waste plastics.

[0094] Embodiment 2

[0095] Referring to Figure 1 , on the basis of the first embodiment, the present embodiment further provides a super-fast analysis system for the components of mixed waste plastics, comprising:

[0096] The collection module acquires image information of a target area and pre-processes the same, and performs anchor frame on the pre-processed image information to obtain waste fragment position information and split the picture, and performs optimal path planning.

[0097] The scanning module uses an infrared scanning module to move based on the optimal path, performs infrared scanning, and analyzes the components of the waste fragments according to the spectral information.

[0098] The calculation module uses a convolutional neural network model to predict the optimal Canny edge detection operator threshold value, performs edge detection and contour extraction, and calculates the area of the waste fragments.

[0099] The analysis module analyzes the components of the mixed waste plastics based on the waste fragment component information and the waste fragment area information.

[0100] The embodiment further provides a computer device suitable for the case of the mixed waste plastic component ultrafast analysis method, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the mixed waste plastic component ultrafast analysis method proposed in the above embodiment.

[0101] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0102] The embodiment further provides a storage medium having a computer program stored thereon, and the program is executed by a processor to realize the mixed waste plastic component ultrafast analysis method proposed in the above embodiment.

[0103] The storage medium proposed in the embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept, and the technical details not described in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.

Claims

1. A method for ultrafast analysis of mixed waste plastic constituents, characterized in that: The method comprises the following steps: An infrared scanning module is used to move along the optimal path and perform infrared scanning, and the composition of the waste fragment is analyzed according to spectral information; A convolutional neural network model is used to predict the optimal Canny edge detection operator threshold value, and edge detection and contour extraction are performed to calculate the area of the waste fragment; Based on the composition information and the area information of the waste fragment, the composition of the mixed waste plastic is analyzed. The optimal path planning comprises: A dynamic programming algorithm and a greedy algorithm are combined to calculate the shortest path for traversing all positions, which is used as the optimal path; An infrared scanning module is used to move along the optimal path and perform infrared scanning, and the composition of the waste fragment is analyzed according to spectral information; The threshold value prediction comprises: A convolutional neural network model is trained to predict the optimal threshold value of the Canny edge detection operator corresponding to the image; A Canny operator and the predicted optimal threshold value are used to perform edge detection on the segmented image; Dilation and erosion operations are performed on the detected edge information to obtain the closed contour line of each fragment; Based on the closed contour, the contour area is calculated to obtain the area information of the fragment; The Canny operator is used to perform convolution operation on the image information to obtain gradient information; The expressions of the Canny operator in the x and y directions are as follows: The preprocessing comprises the following steps: An RGB image of the mixed waste plastic sample is obtained, and the RGB image is completely corresponding to the target region range through image cropping, which comprises the following steps: where S x for computing the image x-direction pixel gradient matrix G x , y for computing the image y-direction pixel gradient matrix G y , I is the grayscale image matrix, and * denotes the cross-correlation computation; Computing the gradient strength matrix G xy : where g x (i,j) is the element of G x in the i-th row and j-th column of the matrix, g y (i,j) is the element of G y in the i-th row and j-th column of the matrix, g xy (i,j) is the element of G xy in the i-th row and j-th column of the matrix.

2. The method of claim 1, wherein the method is characterized by: An image of the target region is captured to obtain an RGB image of the mixed waste plastic sample; The RGB image is imported into an image processing system; The image is displayed in the image processing system, and a deformable quadrilateral frame is generated on the image; The shape and position of the quadrilateral frame are adjusted by moving the four vertices of the quadrilateral frame to cover the target region in the image; The image is cropped through the quadrilateral frame to obtain a cropped image; The shape of the cropped image is corrected to completely correspond to the original rectangular shape of the target region. The waste fragment position information acquisition comprises the following steps:

3. The method of claim 2, wherein the method is characterized by: A data set composed of RGB images of mixed waste samples is established, and waste plastic fragments in the images are labeled; A YOLOv5 model is trained using the labeled data set; The RGB image is anchored based on the trained YOLOv5 model to determine the waste fragment position information. The analysis of the composition of the mixed waste plastic is based on the fragment composition information and the area information.

4. The method of claim 3, wherein the method is characterized by: The method comprises the following steps:

5. A mixed waste plastic component ultrafast analysis system based on the mixed waste plastic component ultrafast analysis method according to any one of claims 1 to 4, characterized in that: An image collection module is used to obtain image information of a target region and perform preprocessing on the image information, anchor the preprocessed image information, obtain waste fragment position information, and segment the image to plan an optimal path. ​ The scanning module adopts an infrared scanning module to move based on the optimal path and perform infrared scanning, and analyze the composition of the waste fragment according to spectral information; The computing module uses a convolutional neural network model to predict the optimal Canny edge detection operator threshold, and performs edge detection and contour extraction to calculate the area of the waste fragment; The analysis module analyzes the composition of the mixed waste plastic based on the composition information and the area information of the waste fragment. 6.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the mixed waste plastic composition ultrafast analysis method in any one of claims 1-4.

7. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the mixed waste plastic composition ultrafast analysis method in any one of claims 1-4.

Images