Image recognition method, material sorting method, recognition device and sorting equipment

Abstract

This disclosure relates to the field of ore sorting technology, specifically to image recognition methods, material sorting methods, recognition devices, and sorting equipment. The image recognition method includes: acquiring a ray image of the material using a ray device and a color-sorted image of the material using a color-sorting camera; determining the ray contour of the material based on the ray image and the color-sorted contour of the material based on the color-sorted image; determining the motion state of the material relative to the conveying device based on the ray contour and the color-sorted contour; if the motion state is motion, then: determining a candidate contour in the color-sorted image corresponding to the ray contour based on the ray contour; and adjusting the candidate contour based on its position in the color-sorted image to obtain a first target contour of the material in the color-sorted image. This method can effectively improve the accuracy of contour matching between the ray image and the color-sorted image, thereby improving the accuracy of image recognition. In the material identification and sorting process, it improves the accuracy of material identification, thus improving the accuracy of sorting.

Description

Technical Field

[0001] This disclosure relates to the field of ore sorting technology, specifically to image recognition methods, material sorting methods, recognition devices, and sorting equipment. Background Technology

[0002] Coal, spodumene, and silica are widely used in current industrial production. However, ores such as coal, spodumene, and silica typically contain large amounts of gangue and impurities after mining, requiring sorting to improve ore quality, reduce transportation costs, and mitigate environmental pollution. Ore sorting equipment is widely used to meet this need. Ore sorting generally involves conveying the ore using a conveyor belt, with color sorting cameras and X-ray devices installed along the conveying path to identify and detect the ore, followed by sorting by a sorting device. However, existing ore sorting equipment has certain technical bottlenecks. Because the ore being inspected may move relative to the conveyor, the outlines of the ore collected and detected by the color sorting camera and X-ray device may not match. This leads to inaccurate images captured during ore identification, reducing the accuracy of ore identification and increasing the likelihood of incorrect or missed sorting, ultimately lowering the overall ore sorting accuracy. Summary of the Invention

[0003] To overcome the problems existing in related technologies, an exemplary embodiment of this disclosure provides an image recognition method applied to a material sorting device. The material sorting device includes: a conveying device for conveying materials, a ray device disposed above the conveying device, and a color sorting camera disposed above the conveying device and downstream of the ray device. The image recognition method includes: acquiring a ray image of the material through the ray device and acquiring a color sorting image of the material through the color sorting camera; determining a ray contour of the material based on the ray image and determining a color sorting contour of the material based on the color sorting image; determining the motion state of the material relative to the conveying device based on the ray contour and the color sorting contour; if the motion state is motion, then: determining a candidate contour in the color sorting image corresponding to the ray contour based on the ray contour; and adjusting the candidate contour based on the position of the candidate contour in the color sorting image to obtain a first target contour of the material in the color sorting image.

[0004] In some embodiments, adjusting the candidate contour based on its position in the color sorting image to obtain a first target contour of the material in the color sorting image includes: determining a background region in the candidate contour based on its position and a background image in the color sorting image; and determining the first target contour based on the candidate contour and the background region.

[0005] In some embodiments, determining the first target contour based on the candidate contour and the background region includes: determining a first centroid based on the candidate contour; determining a second centroid corresponding to a non-background region in the candidate contour based on the background region; determining an adjustment displacement based on the first centroid and the second centroid; and adjusting the candidate contour based on the adjustment displacement to obtain the first target contour.

[0006] In some embodiments, adjusting the candidate contour based on the adjustment displacement to obtain the first target contour includes: obtaining the candidate position of the candidate contour based on the adjustment displacement; adjusting the candidate position and / or the size of the candidate contour to obtain a plurality of intermediate contours; and selecting the intermediate contour with the smallest area as the first target contour based on the area of ​​the background image within each intermediate contour.

[0007] In some embodiments, determining the motion state of the material relative to the conveying device based on the ray profile and the color sorting profile includes: determining a reference profile in the ray image corresponding to the color sorting profile based on the color sorting profile; and determining the motion state of the material relative to the conveying device based on the reference profile and the ray profile.

[0008] In some embodiments, determining the motion state of the material relative to the conveying device based on the reference profile and the ray profile includes: determining the motion state of the material relative to the conveying device based on the distance between the theoretical centroid of the reference profile and the actual centroid of the ray profile; or, determining the motion state of the material relative to the conveying device based on the intersection-union ratio of the reference profile and the ray profile; or, determining the motion state of the material relative to the conveying device based on the axis length and direction of the ellipse fitted by the reference profile or the ray profile.

[0009] In some embodiments, the image recognition method further includes: if the motion state is stationary, then: based on the ray profile, determining a second target profile of the material in the color sorting image that corresponds to the ray profile.

[0010] Secondly, this disclosure also provides a material sorting method applied to a material sorting device, the material sorting device comprising: a conveying device for conveying materials, an ray device disposed above the conveying device, a color sorting camera disposed above the conveying device and downstream of the ray device, and a sorting device disposed downstream of the conveying device; determining a target image of the material in a color sorting image based on a target contour, wherein the target contour is obtained by an image recognition method as described in the first aspect; determining the category of the material based on the target image; and sorting the material by the sorting device based on the category.

[0011] In some embodiments, determining the category of the material based on the target image includes: determining the category based on the target image using a classification model, wherein the classification model is trained using multiple color-sorted samples, and the color-sorted samples are obtained using the image recognition method as described in the first aspect.

[0012] Thirdly, this disclosure also provides an identification device for material sorting, the identification device comprising: an X-ray device for acquiring X-ray images of the material; and a color sorting camera disposed downstream of the X-ray device for acquiring color sorting images of the material; wherein the identification device identifies the material using the image recognition method as described in the first aspect.

[0013] Fourthly, this disclosure also provides a sorting device for sorting materials, wherein the sorting device includes: a conveying device for conveying the materials; an identification device as described in the third aspect for identifying the materials; and a sorting device for sorting the materials according to the identification result of the identification device.

[0014] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure.

[0015] According to the image recognition method provided in this disclosure, contour recognition and registration can be performed on X-ray images acquired by the X-ray device of a material sorting equipment and color sorting images acquired by a color sorting camera, thereby obtaining the target contour in the color sorting image and making the target contour closer to the actual contour of the material. This disclosure effectively improves the accuracy of contour matching between X-ray images and color sorting images, thus improving the accuracy of image recognition. The image recognition method provided in this disclosure can effectively improve the identification accuracy of materials in subsequent material identification and sorting processes, thereby improving the accuracy of material sorting by the sorting equipment. Attached Figure Description

[0016] The invention can be better understood by describing exemplary embodiments of the invention in conjunction with the accompanying drawings, in which:

[0017] Figure 1 This is a flowchart illustrating an image recognition method according to an exemplary embodiment of the present disclosure;

[0018] Figure 2 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0019] Figure 3 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0020] Figure 4 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0021] Figure 5 This is a schematic diagram illustrating a candidate contour adjustment process according to an exemplary embodiment of the present disclosure;

[0022] Figure 6 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0023] Figure 7 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0024] Figure 8 This is a flowchart illustrating an image recognition method according to another exemplary embodiment of the present disclosure;

[0025] Figure 9 This is a flowchart illustrating a material sorting method according to another exemplary embodiment of the present disclosure;

[0026] Figure 10 This is a flowchart illustrating a material sorting method according to another exemplary embodiment of the present disclosure. Detailed Implementation

[0027] The following describes specific embodiments of the present invention. It should be noted that, in order to provide a concise description, this specification cannot exhaustively describe all features of the actual embodiments. It should be understood that, in the actual implementation of any embodiment, just as in any engineering or design project, various specific decisions are often made to achieve the developer's specific goals and to meet system-related or business-related constraints, and this can change from one embodiment to another. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this invention, some design, manufacturing, or production modifications based on the technical content disclosed herein are merely conventional technical means and should not be construed as insufficient content of this disclosure.

[0028] Unless otherwise defined, the technical or scientific terms used in the claims and description shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in the patent application description and claims of this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. The terms "an" or "a" and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "comprising" or "including" and similar terms mean that the element or object preceding "comprising" or "including" encompasses the element or object listed following "comprising" or "including" and its equivalents, and do not exclude other elements or objects. The terms "connected" or "linked" and similar terms are not limited to physical or mechanical connections, nor are they limited to direct or indirect connections.

[0029] Material sorting equipment is used to identify and sort materials, thereby classifying them and collecting different categories separately. Typically, materials are transported via a conveyor system, with a X-ray device and a color sorting camera positioned above the conveyor to capture X-ray images and color sorting images of the materials, respectively. The sorting equipment can determine the material category based on these images and ultimately separate the different categories. Specifically, the material can be ore, which may include coal and gangue. Material sorting equipment can identify the ore, separating coal from gangue, thus separating impurities from the coal, resulting in higher purity and better quality coal. In determining the material category using X-ray and color sorting images, X-ray images provide clearer and more accurate contour information, while color sorting images can display the texture, color, and other characteristics of the material surface. Therefore, determining material categories and performing sorting by recognizing X-ray images and color sorting images requires extracting the contour information from the X-ray images, extracting the material contours from the color sorting images, and extracting features within the material contours in the color sorting images to determine the material category. However, due to vibrations and speed fluctuations in the conveyor, materials may move relative to the conveyor during transport. This relative movement can cause misalignment between the material contours determined in the X-ray images and the image areas containing the material features in the color sorting images, resulting in inaccurate extraction. In this situation, the mismatch between the material contours acquired and detected by the color sorting camera and the X-ray device easily leads to errors in material identification, resulting in low accuracy and precision, and increasing the likelihood of incorrect or missed sorting, thus reducing the overall accuracy of material sorting.

[0030] To overcome the problems existing in related technologies, such as Figure 1As shown, an exemplary embodiment of this disclosure provides an image recognition method applied to a material sorting device. The material sorting device includes: a conveying device for transporting materials, a ray device disposed above the conveying device, and a color sorting camera disposed above the conveying device and downstream of the ray device. The conveying device of the material sorting device can be a belt conveyor, chain conveyor, or roller conveyor, etc. The materials can be conveyed to the sorting device of the material sorting device as the conveying device moves, where different types of materials are separated and collected. Both the ray device and the color sorting camera can be disposed above the conveying device, so as to be able to photograph the materials located on the top surface of the conveying device and moving with the movement of the conveying device. Both the ray device and the color sorting camera are disposed above the conveying device, and the color sorting camera can be located downstream of the ray device. The sorting device transports materials through the conveying device. During the process of the materials being transported to the sorting device by the conveying device, images can be acquired sequentially by the ray device and the color sorting camera according to the image recognition method, and then the images are processed to finally obtain relatively accurate contour information of the materials. The image recognition method may include steps S110 to S150.

[0031] Step S110: Acquire a X-ray image of the material using a X-ray device, and acquire a color-sorted image of the material using a color-sorting camera. The X-ray device is located above the conveyor and emits rays, such as X-rays, onto the surface of the conveyor. The material can be located on the surface of the conveyor and moves along the direction of movement of the conveyor. When the material moves to the image acquisition area corresponding to the X-ray device, the X-ray device emits rays onto the surface of the conveyor, acquiring data from the image acquisition area, thus obtaining a X-ray image corresponding to that area. The X-ray image can at least include the outline information of the material. The color-sorting camera can be an industrial camera, such as a CCD camera, an area scan camera, or a line scan camera. The color-sorting camera can also be located above the conveyor and downstream of the X-ray device. Therefore, after the material passes through the image acquisition area corresponding to the X-ray device, it moves to the image acquisition area corresponding to the color-sorting camera as the material moves with the conveyor. The color-sorting camera can capture an image of its corresponding image acquisition area as a color-sorted image. The color-sorted image can include information about the material itself, where the material is the material acquired in the corresponding X-ray image. In addition, the color-sorted image can also include information such as the color of the material surface.

[0032] Step S120: Determine the ray contour of the material based on the ray image, and determine the color sorting contour of the material based on the color sorting image. Based on the ray image, the ray contour of the material acquired in the ray image can be determined using methods such as contour extraction algorithms, deep learning models, or edge detection. The ray contour is the contour corresponding to the material in the ray image captured by the ray equipment. Based on the color sorting image, the color sorting contour in the color sorting image can be determined using methods such as contour extraction algorithms, deep learning models, or edge detection. The color sorting contour is the contour corresponding to the material in the color sorting image captured by the color sorting camera.

[0033] Step S130: Determine the motion state of the material relative to the conveying device based on the ray profile and color sorting profile. Since the distance between the ray device and the color sorting camera is constant, and the time interval between each ray image and the color sorting image is the same, for the material on the conveying device, when it remains relatively stationary, its pose relative to the conveying device does not change, and the shape and position of its ray profile and color sorting profile should be the same. However, when the material moves relative to the conveying device, its pose may change, resulting in a difference in the shape and / or position of the ray profile and color sorting profile. Specifically, relative motion between the material and the conveying device may be due to the material having acceleration, causing its position on the conveying device to change as it moves, resulting in different positions for the ray profile and color sorting profile. Relative motion may also be due to the material rotating, causing a change in its posture on the conveying device, resulting in a difference between the ray profile and color sorting profile. Therefore, the relative motion state of the material relative to the conveying device can be determined as stationary or in motion based on the ray profile and color sorting profile.

[0034] If the motion state is motion, then steps S140 and S150 can be executed.

[0035] Step S140: Based on the ray profile, determine the candidate profile corresponding to the ray profile in the color sorting image. When the motion state is in motion, the material may have a certain acceleration, causing the material's moving speed to differ from the conveying speed of the conveyor, or the material may rotate during transmission, causing a change in the material's posture. Since the color sorting profile is obtained by extracting the profile from the color sorting image, while the ray profile is obtained by extracting the profile from the ray image, it is difficult to compare the color sorting profile and the ray profile in the same dimension. Therefore, the candidate profile corresponding to the ray profile in the color sorting image can be determined based on the information of the ray profile, the distance between the ray device and the color sorting camera, and the conveying speed of the material transported by the conveyor. For the candidate profile, after the material is captured by the ray device, assuming that the material and the conveyor belt remain relatively stationary, the profile of the material captured by the color sorting camera when the material moves to the color sorting camera is the candidate profile. Specifically, since the time interval for the ray device to capture the ray image, the time interval for the color sorting camera to capture the color sorting image, the distance between the ray device and the color sorting camera, and the conveying speed of the conveyor belt are all preset fixed parameters. Therefore, based on the ray profile, the distance between the ray device and the color sorting camera, and the conveyor belt speed, the candidate profile corresponding to the ray profile in the color sorting image can be determined.

[0036] Step S150: Based on the position of the candidate contour in the color sorting image, adjust the candidate contour to obtain the first target contour of the material in the color sorting image. Since the relative motion between the material and the conveying device is dynamic, the candidate contour and the actual contour position of the material in the color sorting image differ. Therefore, it is necessary to adjust the candidate contour according to its position in the color sorting image to determine the first target contour in the color sorting image. Adjusting the candidate contour can be done by directly translating, rotating, or scaling the candidate contour, or by adjusting the position of each contour point of the candidate contour to obtain an adjusted candidate contour, which is then used as the first contour of the color sorting image. Alternatively, adjusting the candidate contour can be done by adjusting the position information of each contour point of the candidate contour to obtain a new contour as the first target contour in the color sorting image without changing the original definition of the candidate contour. Specifically, the contour points of the candidate contour can be moved according to their positions to determine the new adjusted position information of each contour point, and the new contour formed by the adjusted contour points is used as the first target contour. The first target contour can be the actual contour of the material in the color sorting image.

[0037] According to the image recognition method provided in this embodiment, by comparing and analyzing the X-ray image and color sorting image of the material, contour recognition and registration can be performed on the X-ray image acquired by the X-ray device of the material sorting equipment and the color sorting image acquired by the color sorting camera. By obtaining the target contour in the color sorting image through this embodiment, the target contour can be made closer to the actual contour of the material. This avoids the problem of excessive difference between the color sorting image obtained from the target contour and the actual color sorting image corresponding to the material due to relative movement between the material and the conveying device. Even if the material moves or rotates during transmission, the actual contour of the material in the color sorting image can be accurately restored, thus enabling more accurate acquisition of material characteristics and higher recognition accuracy of the material sorting equipment, avoiding misidentification and missed identification. The image recognition method provided in this embodiment can effectively improve the accuracy of contour matching between X-ray images and color sorting images, improve the accuracy of image recognition, and thus effectively improve the accuracy of material identification and sorting. In subsequent material identification and sorting processes, it can effectively improve the identification accuracy of materials, thereby improving the accuracy of material sorting by the sorting equipment.

[0038] In some embodiments, such as Figure 2 As shown, step S150, which adjusts the candidate contour based on its position in the color sorting image to obtain the first target contour of the material in the color sorting image, may include steps S151 and S152.

[0039] Step S151: Based on the position of the candidate contour and the background image in the color sorting image, determine the background region in the candidate contour. The color sorting image may include a background image and the area where the material is located. The background image is the area outside the material captured in the color sorting image, i.e., the upper surface of the conveying device. Since the color and shape of the area where the material is located in the color sorting image are not necessarily uniform, while the color of the top surface of the conveying device is uniform and its proportion in the color sorting image is large, making it easy to detect, the background image in the color sorting image can be determined, thereby enabling more accurate image recognition. The background image can be an image of the top surface of the conveying device. Based on the position of the candidate contour in the color sorting image, the area enclosed by the candidate contour in the color sorting image can be determined. Based on the background image in the color sorting image, combined with the position of the candidate contour in the color sorting image, the background region within the area enclosed by the candidate contour can be determined. The background region is the area of ​​the top surface of the conveying device captured within the area enclosed by the candidate contour in the color sorting image.

[0040] Step S152: Determine the first target contour based on the candidate contour and the background area. For the area enclosed by the candidate contour in the color sorting image, the background area corresponds to the top surface of the conveying device, while the non-background area corresponds to the material captured in the color sorting image. Based on the position of the candidate contour in the color sorting image and the portion of the color sorting image it encloses, combined with the background area in the candidate contour determined in step S151, the candidate contour can be adjusted according to the difference between the area enclosed by the candidate contour in the color sorting image and the background area. This allows for the determination of the area corresponding to the material within the area enclosed by the adjusted candidate contour, and the determination of the contour edge of this area, thus determining the first target contour. The first target contour is the actual contour corresponding to the material in the color sorting image.

[0041] According to the material identification method provided in this embodiment, step S151, through background recognition of candidate contours, accurately distinguishes the background area corresponding to the conveying device and the area corresponding to the material in the image. This eliminates interference from the background area, making the extracted first target contour more accurate and closer to the actual contour of the material in the color sorting image. This effectively avoids the influence of the background on material identification and sorting during subsequent identification and sorting processes based on the first target contour, thereby improving the accuracy of image recognition and the overall material sorting accuracy of the sorting equipment. Step S152, by determining the first target contour, effectively removes background information through a combination of background area analysis and contour adjustment. It detects and adjusts candidate contours that incorrectly contain background, ensuring that the final contour only surrounds the actual material area and that the identified contour conforms to the actual material boundary, making the final first target contour more accurate. This allows the material sorting equipment to classify materials based on more accurate shape, color, and other features, reducing missorting and improving the accuracy of material sorting.

[0042] In some embodiments, such as Figure 3 As shown, step S152, which determines the first target contour based on the candidate contour and the background region, may include steps S1521 to S1524.

[0043] Step S1521: Determine the first centroid based on the candidate contour. For the candidate contour determined in the preceding steps, the position of the candidate contour can be determined, i.e., the position information of each contour point on the candidate contour in the color sorting image. Based on the position information of the candidate contour and its contour points, the first centroid of the candidate contour can be directly determined by calculating the geometric center. This allows us to determine the position of the first centroid in the color sorting image. Alternatively, the first centroid can be determined using image moments based on the region enclosed by the candidate contour in the color sorting image. Furthermore, the candidate contour can be represented by a parameterized equation, and the first centroid can be calculated using methods such as boundary integrals, thereby determining the position information of the first centroid of the candidate contour.

[0044] Step S1522: Based on the background region, determine the second centroid corresponding to the non-background region in the candidate contour. Based on the background region in the candidate contour, the difference between the region enclosed by the candidate contour and the background region can be calculated to determine the non-background region in the candidate contour. The non-background region can be the region corresponding to the material in the color sorting image enclosed by the candidate contour. Because the material moves relative to the conveying device during its movement, the intersection between the actual area of ​​the material in the color sorting image and the region enclosed by the candidate contour is small. Therefore, the non-background region in the candidate contour is only a part of the actual area corresponding to the material acquired in the color sorting image. The non-background region in the candidate contour can be determined based on the background region, and the second centroid can be determined for the non-background region. The location information of the second centroid of the non-background region can be determined by calculating the contour points of the non-background region after determining its contour. Alternatively, the location information of the second centroid of the non-background region can be determined by binarizing the non-background region and then calculating image moments.

[0045] Step S1523: Determine the adjustment displacement based on the first centroid and the second centroid. Based on the position information of the first centroid and the second centroid determined in steps S1523 and S1524 respectively, the adjustment displacement of the candidate contour can be determined. The adjustment displacement includes the direction and distance of movement for adjusting the candidate contour. Specifically, the adjustment vector from the first centroid to the second centroid can be determined based on the position information of the first centroid and the second centroid in the color sorting image. The direction of the adjustment vector is the direction of the adjustment displacement, and the magnitude of the adjustment displacement can be determined based on the distance of the adjustment vector. Specifically, the magnitude of the adjustment displacement can be twice the distance of the adjustment vector. Since the position of the candidate contour in the color sorting image is the theoretical position of the ray contour in the ray image, and the non-background area is the area occupied by the material in the candidate contour, when the material is stationary relative to the conveying device, the color sorting image enclosed by the candidate contour should be entirely or mostly non-background areas, and the positions of the first centroid of the candidate contour and the second centroid of the non-background area should be close to or coincide with each other. When the material moves relative to the conveying device, the area corresponding to the material in the color sorting image shifts, resulting in a smaller proportion of the non-background area in the color sorting image enclosed by the candidate contour, and a larger shift in the second centroid relative to the first centroid. Therefore, the direction of material movement relative to the candidate contour can be determined based on the adjustment vector and the direction from the first centroid to the second centroid. The distance of material movement relative to the candidate contour can be determined based on the distance from the first centroid to the second centroid.

[0046] Step S1524: Adjust the candidate contour based on the adjustment displacement to obtain the first target contour. Based on the direction and distance of the adjustment displacement determined in step S1523, the candidate contour is adjusted so that the coordinates of each contour point of the candidate contour move along the direction of the adjustment displacement, and the moving distance is the distance of the adjustment displacement, thereby obtaining a new set of contour points. The set of the determined new contour points is defined as the first target contour.

[0047] According to the image recognition method provided in this embodiment, the displacement is calculated and adjusted by determining the first centroid of the candidate contour and the second centroid of the non-background region, so that the candidate contour can more accurately align with the actual position of the material in the color sorting image. The image recognition method provided in this embodiment can effectively correct contour offset caused by the movement of the material relative to the conveying device, improving the matching accuracy of the target contour. Furthermore, by adjusting the candidate contour, it is ensured that the first target contour more closely matches the actual shape of the material, thereby optimizing the material recognition effect and improving the accuracy and stability of subsequent sorting. This method enhances the accuracy of material contour extraction in the color sorting image, further improving the overall recognition accuracy and sorting efficiency of the material sorting equipment.

[0048] In some embodiments, such as Figure 4 As shown, step S1524, which adjusts the candidate contour based on the adjusted displacement to obtain the first target contour, may include steps S15241 to S15243.

[0049] Step S15241: Based on the adjusted displacement, the candidate position of the candidate contour is obtained. The adjustment displacement allows for adjustments to the movement direction and distance of the candidate contour, thereby obtaining its candidate position. The candidate position can be the centroid of the candidate contour. Specifically, for example... Figure 5 As shown, in the image It can be the first centroid, and the corresponding rectangle in the image is the candidate contour corresponding to the ray contour in the color selection image. It can be the second centroid, where, with The region enclosed by the rectangle with centroid is part of the candidate contour region. The rectangle with centroid represents the non-background region within the candidate contour, i.e., the area containing material captured within the candidate contour. This can be determined based on... This involves determining the adjustment displacement, including the direction and magnitude of movement of the candidate contour. The direction of adjustment is used as the direction of displacement, and the candidate contour can be translated along this direction. The displacement adjustment can be... The first centroid can be moved based on the first centroid, the second centroid, and the adjustment displacement. Thus, the third mass center is determined. The centroid corresponding to the candidate position of the candidate contour. Furthermore, assuming the contour point coordinates of the candidate contour are C, the coordinates of the contour point corresponding to the candidate position after the displacement can be calculated based on the adjusted displacement. .

[0050] Step S15242: Adjust the candidate position and / or the size of the candidate contour to obtain multiple intermediate contours. Since the adjusted candidate contour may be the contour corresponding to the material, or it may be close to the contour corresponding to the material, after determining the candidate position of the candidate contour, the size of the candidate position or candidate contour can be fine-tuned according to the adjustment displacement to obtain multiple intermediate contours. Because the material may move and rotate relative to the conveying device, and the material may change its actual contour shape in the color sorting image due to its rotation or tumbling, the candidate contour may move, rotate, and deform relative to the actual contour of the material. Multiple intermediate contours can be obtained by adjusting the overall size of the candidate contour. Alternatively, the candidate contour can be translated or rotated, and multiple intermediate contours can be obtained by adjusting the candidate position of the candidate contour. Specifically, the adjustment range of the candidate contour can be predetermined, such as setting the adjustment range of each contour point of the candidate contour to a range of 5 pixels up, down, left, and right, so that the contour points of the candidate contour are translated and rotated within this adjustment range, thereby achieving the adjustment of the overall size and / or candidate position of the candidate contour and determining multiple intermediate contours. Furthermore, the overall adjustment of all contour points of the candidate contour can be performed based on the direction and magnitude of the displacement adjustment. The position and size of the candidate contour can be adjusted according to the material movement trend based on the direction and magnitude of the displacement adjustment, thereby obtaining multiple intermediate contours. Specifically, the adjustment range of each contour point of the candidate contour can be set to a range of 5 pixels vertically, horizontally, and vertically. Within this range, the contour points of the candidate contour are synchronously translated, causing the candidate contour to be translated as a whole, thus adjusting the candidate's position. Alternatively, the contour points of the candidate contour can be adjusted independently, thereby adjusting the size of the candidate contour.

[0051] Step S15243: Based on the area of ​​the background image within each intermediate contour, the intermediate contour with the smallest area is selected as the first target contour. For the multiple intermediate contours determined by step S15243, the area of ​​the background image in each intermediate contour can be determined. By comparing the areas of the background images of all intermediate contours, the intermediate contour with the smallest background image area can be selected as the first target contour. Since the material can move, rotate, and tumble relative to the conveyor belt, there may be some deviation between the intermediate contour and the actual material contour, resulting in a portion of the background image within the area enclosed by the intermediate contour. According to step S15243, the area of ​​the background image portion in each intermediate contour can be compared, thereby determining the intermediate contour with the smallest background image area. The area enclosed by this intermediate contour has the largest area occupied by the material, thus determining that this intermediate contour is closest to the actual color sorting contour of the material in the color sorting image. Therefore, this intermediate contour can be selected as the first target contour.

[0052] According to the image recognition method provided in this embodiment, adjusting the position and size of the candidate contour based on displacement adjustment enables the first target contour obtained after adjustment to more accurately match the actual contour of the material in the color sorting image. By adjusting the position and fine-tuning the size of the candidate contour, it is possible to effectively compensate for the movement, rotation, and deformation that may occur during the material's transport, improving the stability and robustness of the matching between the material's color sorting contour and the ray contour. This ensures that in the subsequent process of extracting the corresponding area of ​​the material in the color sorting image based on the first target contour, the extracted color sorting image is closer to the color sorting image corresponding to the actual material contour, thereby improving the accuracy of subsequent material recognition. In addition, by comparing the area of ​​the background region in multiple intermediate contours, the first target contour that best matches the actual material contour can be selected, further improving the accuracy of image recognition. This method can significantly improve the image recognition and material sorting accuracy of the material sorting equipment, enhancing the accuracy and reliability of sorting.

[0053] In some embodiments, such as Figure 6 As shown, step S130, which determines the motion state of the material relative to the conveying device based on the ray profile and color sorting profile, may include steps S131 and S132.

[0054] Step S131: Based on the color sorting profile, determine the reference profile corresponding to the color sorting profile in the X-ray image. By using the color sorting profile corresponding to the material in the color sorting image, and considering the distance between the X-ray device and the color sorting camera, as well as the conveyor speed of the material, the reference profile corresponding to the color sorting profile in the X-ray image can be determined. In determining the relative motion state between the material and the conveyor belt based on the color sorting profile and the X-ray profile, it is necessary to evaluate the color sorting profile and the X-ray profile in the same dimension. Therefore, step S131 can be used to determine the reference profile corresponding to the color sorting profile in the X-ray image. Specifically, for the reference profile, assuming that the material and the conveyor belt remain relatively stationary, the color sorting profile of the material acquired by the color sorting camera is used as a reference, and the color sorting profile is mapped onto the X-ray image as the reference profile. Since the time interval for the X-ray device to acquire the X-ray image, the time interval for the color sorting camera to acquire the color sorting image, the distance between the X-ray device and the color sorting camera, and the conveyor speed are all preset fixed parameters, the reference profile corresponding to the color sorting profile in the X-ray image can be determined based on the color sorting profile, the distance between the X-ray device and the color sorting camera, and the conveyor speed.

[0055] Step S132: Determine the motion state of the material relative to the conveying device based on the reference profile and the ray profile. The motion state of the material relative to the conveying device can be determined based on the offset of the reference profile relative to the ray profile. Specifically, the reference profile and the ray profile can be compared, and the relative motion state between the material and the conveying device can be determined based on the degree of difference between them. Since the material may move or rotate in any direction relative to the conveying device during its movement, the deviation between the reference profile and the ray profile may include positional deviation and rotational deviation. The relative motion state between the material and the conveying device can be determined based on the deviation between the reference profile and the ray profile. Specifically, the relative motion state between the material and the conveying device can be determined based on the difference between the reference profile and the ray profile. If the reference profile and the ray profile basically coincide, it indicates that there is no significant relative motion between the material and the conveying device from the acquisition of the ray image to the acquisition of the color sorting image, and the material can be considered to remain stable. If there is a significant offset between the theoretical ray profile and the actual ray profile, it indicates that there is relative motion between the material and the conveying device.

[0056] According to the image recognition method provided in this embodiment, a reference contour corresponding to the color sorting contour in the ray image is determined based on the color sorting contour, and the motion state of the material relative to the conveying device is determined based on the difference between the reference contour and the ray contour. The motion state of the material relative to the conveying device is accurately determined so that different recognition methods can be adopted for the material according to different motion states. This saves the computing power of the sorting equipment that performs the image recognition method, and improves the contour matching accuracy between the ray image and the color sorting image in subsequent method steps. This improves the stability and robustness of the matching between the color sorting image and the ray image, and enhances the accuracy and reliability of material recognition and sorting.

[0057] In some embodiments, such as Figure 7 As shown, step S132, which determines the motion state of the material relative to the conveying device based on the reference profile and the ray profile, may include: step S1321, or step S1322, or step S1323.

[0058] Step S1321: Determine the motion state of the material relative to the conveying device based on the distance between the theoretical centroid of the reference contour and the actual centroid of the ray contour. The theoretical centroid of the reference contour can be determined based on the reference contour, and the actual centroid of the ray contour can be determined based on the ray contour. The theoretical and actual centroids can be determined directly by calculating the position information of their corresponding contour points, or they can be determined using methods such as image moments or boundary integrals based on the reference contour or ray contour. Based on the determined position information of the theoretical and actual centroids in the ray image, the distance between the theoretical and actual centroids can be determined. A distance threshold can be preset to determine the motion state of the material relative to the conveying device based on the distance between the theoretical and actual centroids. Specifically, when the distance between the theoretical and actual centroids is greater than or equal to the distance threshold, it indicates that the material's offset is large, and it can be determined that the material is moving relative to the conveying device. When the distance between the theoretical and actual centroids is less than the distance threshold, it can be considered that the material has only experienced some minor disturbances, such as minor vibrations of the conveying mechanism or minor material offsets caused by the material's own shape. In this situation, the material offset can be considered small, and the material can be considered stationary relative to the conveying device.

[0059] Step S1322: Determine the motion state of the material relative to the conveying device based on the intersection-union ratio (IU / U) of the reference contour and the ray contour. The reference contour and the area of ​​the ray image it encloses can be determined, thus determining the area corresponding to the reference contour and its position in the ray image. Similarly, the ray contour and the area of ​​the ray image it encloses can be determined, thus determining the area corresponding to the ray contour and its position in the ray image. The area of ​​the intersection of the reference contour and the ray contour, and the area of ​​the union of the reference contour and the ray contour, can be determined separately to determine the IU / U ratio. The closer the IU / U ratio is to 1, the higher the overlap between the reference contour and the ray contour. When the IU / U ratio is 1, it indicates that the reference contour and the ray contour are completely identical, and it can be determined that the material has not undergone relative movement with the conveying device, therefore the motion state of the material relative to the conveying device can be determined to be stationary. A threshold IU / U ratio can be set. When the IU / U ratio is greater than or equal to the threshold, only some negligible slight offset may have occurred between the material and the conveying device, and the motion state of the material relative to the conveying device can be determined to be stationary. When the crossover ratio is less than the crossover ratio threshold, the offset between the material and the conveying device is large, and the motion state of the material relative to the conveying device can be determined as motion.

[0060] Step S1323: Determine the motion state of the material relative to the conveying device based on the axis length and direction of the ellipse fitted by the reference contour or ray contour. For cases where the reference contour and ray contour have an inclusion relationship, the reference contour or ray contour can first be fitted into an ellipse. Specifically, in the case of an inclusion relationship, the reference contour and its corresponding reference contour region contain the ray contour and its corresponding ray contour region; or the ray contour and its corresponding ray contour region contain the reference contour and its corresponding reference contour region. Since some materials move at high speeds on the conveying device, the reference contour or ray contour may experience some offset or deformation. In such cases, the larger contour of the reference contour or ray contour can be fitted into an ellipse. The axis length and direction of the ellipse are determined. According to a preset threshold, if the axis length of the ellipse is less than the preset threshold and the direction of the ellipse meets the preset requirements, the motion state of the material relative to the conveying device can be determined to be stationary. Conversely, if the axis length of the ellipse is greater than or equal to the preset threshold and the direction of the ellipse does not meet the preset requirements, it indicates that the material has undergone a significant offset relative to the conveying device, thus determining the motion state of the material relative to the conveying device as motion.

[0061] Furthermore, in some embodiments, if the reference profile and the ray profile do not intersect, it can be directly determined that the current material does not exist, so that the subsequent sorting device does not perform any sorting operation on the material, thereby reducing the false recognition rate and improving the accuracy of material identification and sorting.

[0062] According to the image recognition method provided in this embodiment, the motion state of the material relative to the conveying device can be determined based on the relationship between the reference contour and the ray contour, enabling a more accurate determination of the relative motion state between the material and the conveying device. The material motion state can be determined using various parameters such as the centroid offset between the reference contour and the ray contour, the intersection-union ratio of the reference contour and the ray contour, or the axis length and direction of the ellipse fitted by the reference contour or the ray contour. This more accurate determination of the relative motion state between the material and the conveying device allows for the execution of different recognition operations based on different relative motion states, thereby saving the computing power of the sorting equipment, improving the efficiency of material recognition and sorting, and ensuring the accuracy of material sorting.

[0063] In some embodiments, such as Figure 8As shown, the image recognition method may further include: if the motion state is stationary, then step S160 is executed to determine the second target contour of the material in the color sorting image corresponding to the ray contour, based on the ray contour. When the motion state is stationary, it can be determined that the material's moving speed is equal to the conveying speed of the conveying device, and the material's moving direction is consistent with the conveying direction of the conveying device, and the material does not rotate. For the case of stationary motion, step S160 can be executed to directly determine the contour corresponding to the ray contour in the color sorting image based on the ray contour, and according to the distance between the ray device and the color sorting camera, and the conveying speed of the conveying device, and use it as the second target contour. The second target contour can be the actual contour corresponding to the material in the color sorting image, and the corresponding area in the color sorting image enclosed by the second target contour is the area where the material is located.

[0064] According to the image recognition method provided in this embodiment, when the material is stationary, the second target contour corresponding to the material in the color sorting image can be directly determined based on the ray contour. This allows the sorting equipment to perform different recognition operations according to different relative motion states between the material and the conveying device, thereby saving the computing power of the sorting equipment, improving the real-time processing capability of the sorting equipment, and increasing the efficiency of material recognition and sorting.

[0065] Based on the same inventive concept, such as Figure 9 As shown, this disclosure also provides a material sorting method applied to a material sorting device. The material sorting device includes: a conveying device for conveying materials, an X-ray device disposed above the conveying device, a color sorting camera disposed above the conveying device and downstream of the X-ray device, and a sorting device disposed downstream of the conveying device.

[0066] Step S210: Based on the target contour, determine the target image of the material in the color sorting image, wherein the target contour is obtained by the image recognition method as described in any of the foregoing embodiments. The target contour may include, based on a first or second target contour determined by the image recognition method, the color sorting image region corresponding to the target contour can be extracted from the color sorting image, and the extracted image is used as the target image. The target image may include the color sorting image of the area where the material is located, and may include features such as color and texture of the material captured by the color sorting camera.

[0067] Step S220: Determine the material category based on the target image. Based on the target image, features such as color and texture of the material captured in the image can be extracted. Image recognition technology can be used to identify and process the target image, and the material category can be determined based on these features. Alternatively, the target image can be input into a pre-trained model for material classification and sorting, thereby determining the material category through the model.

[0068] Step S230: Based on category, the materials are sorted using a sorting device. According to step S220, the category of each material conveyed on the conveyor can be determined. Based on the material category, the sorting device separates materials belonging to different categories, thereby achieving material sorting. The sorting device can separate different categories of materials by blowing air onto the materials using a blowing mechanism. Alternatively, the sorting device can push the materials using a pusher mechanism, thereby separating different categories of materials from each other.

[0069] According to the material sorting method provided in this embodiment, the target image of the material in the color sorting image can be determined based on the target contour. That is, the target image is obtained by subtracting the corresponding color sorting image according to the target contour, and this target image is the area actually corresponding to the material in the color sorting image. Through this embodiment, the area where the material is located in the color sorting image can be accurately extracted, and the obtained target image can accurately reflect the color, texture and other surface features of the material, thereby improving the accuracy of material classification. Finally, it effectively improves the accuracy of material sorting during the sorting process by the sorting device.

[0070] In some embodiments, such as Figure 10As shown, step S220, determining the material category based on the target image, may include: step S221, determining the category based on the target image using a classification model, wherein the classification model is trained using multiple color-sorted samples, and the color-sorted samples are obtained using the image recognition method described in the first aspect. The classification model may be a model trained based on algorithms such as decision trees or SVM (Support Vector Machine). Based on the target contour determined by the image recognition method provided in the foregoing embodiments, the target image can be processed to obtain the color, texture, and other features of the material in the target image. The color, texture, and other features of the material in the target image, along with pre-acquired labels, can be used as training inputs, and trained using algorithms such as decision trees or SVM to obtain the classification model. The pre-acquired label corresponding to the features of each target image may be the material category corresponding to that target image. The classification model may be a model trained based on networks such as CNN (Convolutional Neural Network) or ViT (Vision Transformer). Based on the target contour and the determined target image, the target image can be processed to obtain the ROI (Region of Interest) image. Using the ROI image as input, a classification model is trained using a network such as CNN or ViT to obtain a classification model. The trained classification model can then be used for target image recognition. By inputting the target image into the classification model, the model can directly determine the category of the corresponding material in the target image. The material sorting method provided in this embodiment, through a classification model trained on a large number of color-sorting samples, can more accurately identify and sort materials, significantly improving the accuracy and robustness of material classification. Therefore, it effectively improves the accuracy of material sorting.

[0071] Based on the same inventive concept, this disclosure also provides an identification device for material sorting. The identification device may include an X-ray device and a color sorting camera. The identification device can be disposed above a conveying device, enabling it to identify the materials being transported on the conveying device, thereby acquiring an image of the material and identifying the material based on the image to determine its classification. This facilitates subsequent separation of different categories of materials by a sorting device, achieving material sorting. The identification device can identify the materials using the image recognition method described in any of the foregoing embodiments.

[0072] A radiographic device is used to acquire radiographic images of materials. The radiographic device can be positioned above the conveyor. Relative to the color sorter, the radiographic device can be upstream, allowing the material being conveyed to first pass through the image acquisition area of ​​the radiographic device, and then through the image acquisition area of ​​the color sorter. The radiographic device emits rays, such as X-rays, onto the surface of the conveyor. As the material moves with the conveyor to the corresponding image acquisition area of ​​the radiographic device, the emitted rays can be captured, allowing data from that area to be obtained. This results in a radiographic image of that area, which can at least include the outline information of the material.

[0073] A color sorting camera, positioned downstream of the X-ray unit, is used to acquire color-sorted images of materials. The color sorting camera can be placed downstream of the X-ray unit; after the material passes through the image acquisition area of ​​the X-ray unit, it can be conveyed to the image acquisition area of ​​the color sorting camera via a conveyor device, where the color sorting camera acquires the color-sorted image of the material. The color sorting camera can be an industrial camera, such as a CCD camera, an area scan camera, or a line scan camera. The color-sorted image can capture surface features of the material, such as texture and color.

[0074] The identification device provided in this embodiment can acquire X-ray images and color sorting images using an X-ray device and a color sorting camera, respectively. By combining the information from the X-ray images and color sorting images, the material contour can be identified and registered. This ensures that the extracted target contour closely matches the actual shape of the material, improving the accuracy of material identification and enabling precise identification and classification of materials on the conveying device. Furthermore, the identification device provided in this embodiment assists subsequent separation and sorting operations by providing material classification information to the sorting device. This allows the sorting device to more accurately separate and collect different categories of materials based on their type, effectively improving the accuracy and stability of identification and sorting during the material sorting process, while also significantly increasing sorting efficiency.

[0075] Based on the same inventive concept, this disclosure also provides a sorting device for sorting materials, wherein the sorting device may include: a conveying device, an identification device as in any of the foregoing embodiments, and a sorting device.

[0076] A conveying device is used to transport materials. This device can be a belt conveyor or a chain conveyor, etc., capable of transporting materials and allowing them to pass through identification devices for identification and classification. The conveying device ultimately delivers the materials to a sorting device, where they fall from the end of the conveyor and are sorted by the sorting device.

[0077] The identification device can be used to identify materials. It can acquire X-ray images and color-sorted images of the materials using the X-ray device and color-sorting camera in the aforementioned embodiments, respectively. The X-ray contours of the same material in the X-ray image and its color-sorting contours in the color-sorting image are registered to determine the target image. The target image is then identified to determine the category of the material corresponding to it, facilitating subsequent separation and sorting of different categories of materials using a sorting device.

[0078] A sorting device is used to sort materials based on the identification results of an identification device. The sorting device can separate different categories of materials according to the identification structure of the identification device and the different classifications of the materials. The sorting device may include a pusher plate mechanism, which separates and sorts different types of materials by oscillating the pusher plate to strike the material. The pusher plate mechanism may include multiple pusher plates, and one or more pusher plates can be used to perform the sorting operation according to the position of the material falling. The sorting device may also include a blowing mechanism, which separates and sorts different types of materials by blowing gas onto the material through nozzles. The blowing mechanism may include multiple nozzles, and one or more nozzles can be used to perform the sorting operation according to the position of the material falling.

[0079] The sorting equipment provided in this embodiment enables high-precision identification and classification of materials during the material conveying process, and separates materials in real time based on the identification results, thereby significantly improving sorting accuracy and efficiency. The conveying device ensures stable material delivery to the identification and sorting devices. The identification device uses a ray device and a color sorting camera to acquire ray images and color sorting images of the materials, respectively. By accurately registering the ray contours and color sorting contours and extracting the target image, a more accurate material classification is obtained. The sorting device can separate materials according to the material category determined by the identification device, achieving precise sorting of different categories of materials. The sorting equipment provided in this embodiment improves the accuracy of image recognition and further enhances the overall operating efficiency and reliability of the sorting equipment, as well as the sorting accuracy.

[0080] Specifically, the material sorting equipment provided in this disclosure can sort ores. For the separation and sorting of ores, such as the separation and sorting of coal and gangue, a conveying device can transport the mixed coal and gangue to be sorted. When the ore to be sorted on the conveying device passes through the image acquisition area of ​​the identification device, X-ray images and color sorting images of the ore can be captured sequentially by an X-ray device and a color sorting camera. The color sorting image and X-ray image of the ore are then registered according to the image recognition method provided in this disclosure. This allows the identification device to accurately determine the corresponding position of the ore in the color sorting image and X-ray image, extract the corresponding color sorting area of ​​the ore in the color sorting image, and extract the features of the ore in that area to accurately determine whether the ore is coal or gangue. The identified coal or gangue can fall from the end of the conveying device, and the sorting device can separate and collect the identified coal and gangue separately, thereby achieving the separation of coal and gangue from the ore.

[0081] This application uses specific terms to describe embodiments of the application. Terms such as "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of the application. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Furthermore, certain features, structures, or characteristics in one or more embodiments of the application can be appropriately combined.

[0082] In the context of this application, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0083] Similarly, it should be noted that, in order to simplify the description of the present application and thus aid in the understanding of one or more embodiments, the foregoing description of the embodiments of the present application sometimes combines multiple features into a single embodiment, drawing, or description thereof. However, this disclosure method does not imply that the subject matter of the present application requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of the single embodiments disclosed above.

[0084] The basic concepts have been described above. Obviously, for those skilled in the art, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore remain within the spirit and scope of the embodiments of this application.

Claims

1. A method of image recognition applied to a material sorting apparatus, the material sorting apparatus comprising: A conveying device for conveying materials, a radiation device arranged above the conveying device, and a color sorter camera arranged above the conveying device and downstream of the radiation device; The image recognition method comprises: acquiring a radiation image of the material by the radiation device, and acquiring a color sorting image of the material by the color sorter camera; determining a radiation contour of the material based on the radiation image, and determining a color sorting contour of the material based on the color sorting image; determining a motion state of the material relative to the conveying device based on the radiation contour and the color sorting contour; if the motion state is motion, then: determining a candidate contour corresponding to the radiation contour in the color sorting image based on the radiation contour; and, adjusting the candidate contour based on the position of the candidate contour in the color sorting image to obtain a first target contour of the material in the color sorting image; The adjusting of the candidate contour based on the position of the candidate contour in the color sorting image to obtain a first target contour of the material in the color sorting image comprises: determining a background region in the candidate contour based on the position of the candidate contour and a background image in the color sorting image; determining the first target contour based on the candidate contour and the background region; The determination of the first target contour based on the candidate contour and the background region comprises: determining a first centroid based on the candidate contour; determining a second centroid corresponding to a non-background region in the candidate contour based on the background region; determining an adjustment displacement based on the first centroid and the second centroid; adjusting the candidate contour based on the adjustment displacement to obtain the first target contour.

2. The image recognition method of claim 1, wherein The adjusting of the candidate contour based on the adjustment displacement to obtain the first target contour comprises: obtaining a candidate position of the candidate contour based on the adjustment displacement; adjusting the candidate position and / or the size of the candidate contour to obtain a plurality of intermediate contours; selecting the intermediate contour with the smallest area of background image as the first target contour.

3. The image recognition method according to any one of claims 1-2, wherein, The determination of the motion state of the material relative to the conveying device based on the radiation contour and the color sorting contour comprises: determining a reference contour corresponding to the color sorting contour in the radiation image based on the color sorting contour; determining the motion state of the material relative to the conveying device based on the reference contour and the radiation contour.

4. The image recognition method of claim 3, wherein, The determination of the motion state of the material relative to the conveying device based on the reference contour and the radiation contour comprises: determining the motion state of the material relative to the conveying device according to the distance between the theoretical centroid of the reference contour and the actual centroid of the radiation contour; or, determining the motion state of the material relative to the conveying device according to the intersection-over-union of the reference contour and the radiation contour; or, determining the motion state of the material relative to the conveying device according to the axis length and direction of the ellipse fitted by the reference contour or the radiation contour.

5. The image recognition method of claim 1, wherein, The image recognition method further comprises: If the motion state is static, determining a second target contour of the material in the color selection image corresponding to the ray contour based on the ray contour.

6. A method of sorting material for use with a material sorting apparatus, the material sorting apparatus comprising: A conveying device for conveying the material, a ray device arranged above the conveying device, a color selection camera arranged above the conveying device and downstream of the ray device, and a sorting device arranged downstream of the conveying device; Determining a target image of the material in the color selection image based on a target contour, wherein the target contour is obtained by the image recognition method according to any one of claims 1-5; Determining a category of the material based on the target image; Sorting the material based on the category by the sorting device.

7. The method of sorting material of claim 6, wherein, The determining of the category of the material based on the target image comprises: Determining the category according to the target image by a classification model, wherein the classification model is trained by a plurality of color selection samples obtained by the image recognition method according to any one of claims 1-5.

8. An identification device for material sorting, the identification device comprising: a ray device for obtaining a ray image of the material; a color selection camera arranged downstream of the ray device for obtaining a color selection image of the material; wherein the identification device identifies the material by the image recognition method according to any one of claims 1-5.

9. A sorting apparatus for sorting material, wherein, The sorting equipment comprises: a conveying device for conveying the material; the identification device according to claim 8 for identifying the material; a sorting device for sorting the material according to the identification result of the identification device.

Images