An ore foreign matter detection and identification method and an ore foreign matter detection and identification device

Abstract

The embodiment of the application discloses a kind of ore foreign matter detection identification method and ore foreign matter detection identification device, it is related to ore foreign matter detection field, ore foreign matter detection identification method includes camera component and carries out structured light shooting to mineral, collects three-dimensional structure image in mineral;Image capture is carried out to three-dimensional structure image by identification algorithm, the image feature of each mineral is obtained, and image feature is marked to image;Identification algorithm is positioned and identified according to marked image, and first image content is generated according to identification result;Identification algorithm carries out three-dimensional identification judgment according to the image feature in first image content, and gathers the geometric feature in image, identification algorithm carries out foreign matter marking according to geometric feature, generates second image content;Second image content is grabbed according to foreign matter by grabbing device, and foreign matter is marked to mineral by being provided with identification algorithm, can efficiently control the grabbing of foreign matter by grabbing device to realize the classification of ore and foreign matter.

Description

Technical Field

[0001] This application relates to the field of foreign object detection in ores, and more particularly to a method and device for detecting and identifying foreign objects in ores. Background Technology

[0002] With the rapid development and accelerating pace of my country's ore mining industry, safety issues in the ore mining process have become increasingly prominent. During the mining and transportation process, large iron components such as iron plates, reinforcing bars, steel fibers, and shovel teeth can become mixed in with the ore, causing damage to equipment in multiple production stages, including conveying, crushing, and grinding. This severely impacts production efficiency and significantly affects the sustainability and economic benefits of production. Therefore, strengthening the monitoring and diagnosis of ore production and transportation processes, promptly detecting iron components and issuing warnings and sorting measures to avoid or reduce equipment damage in production stages is of paramount importance for safe production throughout the entire ore mining process.

[0003] Existing methods for removing iron parts have the following shortcomings: Due to significant differences in appearance and texture between ore and iron parts, traditional manual removal methods are not only inefficient but also pose safety hazards. These iron parts are typically sorted and removed manually at the production site, next to the conveyor belt, by observation and emergency production line stops. Because manual removal by observation is prone to visual fatigue, it can lead to missed items, causing instability in the entire system's production. Furthermore, the production site environment is harsh, with severe dust and noise pollution; manually inspecting the conveyor belt to remove foreign objects poses serious safety and occupational health hazards.

[0004] Therefore, how to achieve efficient detection and classification of foreign objects in ores is an important technical problem that urgently needs to be solved. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method and device for detecting and identifying foreign objects in ores. By setting an identification algorithm to mark foreign objects in minerals, the method can efficiently control the grasping device to grasp foreign objects in order to classify ores and foreign objects.

[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0007] This application provides a method for detecting and identifying foreign objects in ores, including:

[0008] The camera module uses structured light to capture images of the minerals, collecting three-dimensional structural images of the minerals.

[0009] The recognition algorithm captures the three-dimensional structure image, obtains the image features of each mineral, and marks the image features.

[0010] The recognition algorithm locates and identifies the marked image, and generates the first image content based on the recognition result;

[0011] The recognition algorithm performs three-dimensional recognition based on the image features in the first image content, and collects geometric features in the image. The recognition algorithm marks foreign objects based on the geometric features and generates second image content.

[0012] The grasping device grasps the foreign object based on the content of the second image.

[0013] Optionally, in some embodiments of this application, the camera component includes an industrial camera and a projector. Both the industrial camera and the projector are mounted on the gripping device. The industrial camera captures images of the ore to generate image information. The control terminal vectorizes the image information to generate a vectorized image. The control terminal transmits the vectorized image to the projector. The projector illuminates the ore based on the vectorized image to generate the three-dimensional structure image.

[0014] Optionally, in some embodiments of this application, the three-dimensional structure image is transmitted to the control terminal, the recognition algorithm performs three-dimensional image capture on the three-dimensional structure image, marks the ores and foreign objects in the three-dimensional structure image, and generates a marked image;

[0015] The marked image includes image features of each of the ore and the foreign object.

[0016] Optionally, in some embodiments of this application, the recognition algorithm generates a three-dimensional coordinate system for the marked image interface, and each of the ore and foreign object in the marked image is located by coordinates;

[0017] The identification algorithm obtains the location coordinates of each ore and foreign object and generates the first image content.

[0018] Optionally, in some embodiments of this application, the identification algorithm for obtaining the coordinates of the ore and the foreign object includes:

[0019] The identification algorithm scans images of the ore and the foreign object using the camera component;

[0020] The recognition algorithm performs line fitting on the edge positions of the ore and the foreign object to generate two-dimensional lines;

[0021] The recognition algorithm marks the two-dimensional lines with two-dimensional coordinates.

[0022] The recognition algorithm expresses the two-dimensional lines of the two-dimensional punctuation using a functional algorithm and then makes a judgment:

[0023] When the two-dimensional line cannot be functionalized by the functionalization algorithm, the recognition algorithm marks the mineral corresponding to the two-dimensional line as ore; when the two-dimensional line is functionalized by the functionalization algorithm, the recognition algorithm marks the mineral corresponding to the two-dimensional line as foreign object.

[0024] Optionally, in some embodiments of this application, the recognition algorithm acquires the mineral from multiple angles and fuses the images acquired from multiple angles using coordinate functions to determine geometric features. The recognition algorithm then generates a foreign object marker based on the geometric feature determination to generate the second image content.

[0025] Optionally, in some embodiments of this application, the second image content includes the coordinate values ​​of the marked ore and the coordinate values ​​and function values ​​of the marked foreign object.

[0026] Optionally, in some embodiments of this application, the identification method further includes ore model data, and the identification algorithm performs big data comparison and identification on the content of the first image based on the ore model data. When the mineral matches the ore model data, the mineral is marked as the ore; when the mineral does not match the ore model data, the mineral is marked as the foreign object.

[0027] The grasping device grasps the foreign object according to the foreign object marker.

[0028] This application also provides a foreign object detection and identification device for ores, the identification device being used to perform any of the foreign object detection and identification methods described above.

[0029] Optionally, in some embodiments of this application, the identification device includes a gripping device, which includes an industrial camera and a projector.

[0030] Compared with the prior art, the beneficial effects of this invention are as follows:

[0031] This invention, by incorporating a camera component and a recognition algorithm, can locate and mark the positions of ores and foreign objects within minerals. The recognition algorithm performs image marking and three-dimensional judgment on the minerals, locating and identifying each object within the minerals. Simultaneously, the recognition algorithm uses an edge-forming algorithm on the image through dilatational erosion to generate two-dimensional lines. Based on the line fitting method, it determines whether the object is an ore or a foreign object and marks it accordingly, facilitating the grasping device to grab the foreign object based on the marked information. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 A schematic diagram of the overall structure and flow of the ore foreign object detection and identification method provided in the embodiments of this application;

[0034] Figure 2 This is a schematic diagram of the overall process structure of the recognition algorithm for obtaining coordinates provided in the embodiments of this application. Detailed Implementation

[0035] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this application. It is understood that the accompanying drawings are provided for reference and illustration only, and are not intended to limit this application. The connection relationships shown in the accompanying drawings are only for clear description and do not limit the connection method.

[0036] Specifically, such as Figure 1 As shown in the figure, this application provides a method for detecting and identifying foreign objects in ore. The method is based on an ore foreign object detection and identification device, which includes a robotic arm and a camera assembly, wherein the camera assembly includes an industrial camera and a projector.

[0037] When photographing minerals, the minerals are first illuminated with light of a specific wavelength by a projector in the camera module. The three-dimensional structure and shape of the minerals are calculated by observing the reflection of light by the minerals. Then, an industrial camera is used to capture the images and generate a three-dimensional structural image.

[0038] This three-dimensional structural image includes the basic shape and location of the mineral.

[0039] After the three-dimensional structural image is generated, it is uploaded to a specific control terminal. The control terminal includes a recognition algorithm that can capture the three-dimensional image and generate a corresponding labeled image. The labeling algorithm assigns a digital label to each mineral in the image, and the shape of the mineral corresponding to each number is stored in the image. Since the mineral stacking has a certain degree of complexity, it is necessary to use structured light for three-dimensional capture and three-dimensional labeling of the minerals to facilitate the identification of mineral samples that cannot be found in two-dimensional state.

[0040] The 3D marked image includes markings for ores and foreign objects. The primary purpose of these markings is to locate each mineral, facilitating grasping by the gripping device. However, before the gripping device can grasp the mineral, the recognition algorithm needs to perform depth calculations. This calculation process involves establishing a 3D coordinate system for the marked image, obtaining the corresponding coordinate values, and generating the first image content. This allows the recognition algorithm to accurately locate the mineral sample using 3D coordinates. Specifically:

[0041] The recognition algorithm establishes a spatial rectangular coordinate system, i.e., a three-dimensional coordinate system, on the three-dimensional plane of the marked image. In this embodiment, since the marked image is taken by an industrial camera, the resulting photo is basically rectangular. The marked image obtained after processing the photo under this rectangular structure will also have a rectangular structure. This structure makes it easy to create a rectangular coordinate system. By using one vertex in the photo as the origin, the corresponding three-dimensional coordinate system can be easily created. In the three-dimensional coordinate system, since the mineral occupies a certain area on the plane, when locating the coordinates of the ore and foreign objects, each mineral needs to be marked. The mark in the mineral is the center point of the mineral. The coordinate images are then fused using coordinate functions to perform geometric feature judgment. For regular-shaped objects, the center point of the mineral is the point where the diagonals of the shape intersect. For irregular-shaped objects, the center point of the mineral is the center point automatically generated by the recognition algorithm for the mineral, which facilitates the processing of the mineral.

[0042] In the above process, when the recognition algorithm marks the center point of the mineral, it simultaneously scans the shape of the mineral. This scanning process involves taking pictures of the mineral with an industrial camera, and then performing edge processing on the image using the recognition algorithm. The image is processed using an expansion-corrosion formula to edge the mineral. In order to achieve better edge processing results, a deep convolutional neural network (DCNN) is used to process the image. Alternatively, the CED algorithm can be set to use a top-down reverse thinning method, employing sub-pixel convolution (SPC) to generate clear edges.

[0043] After generating the edge, the recognition algorithm fits the edge to form a two-dimensional line.

[0044] Two-dimensional lines can be marked with two-dimensional coordinates using a two-dimensional coordinate system generated by a recognition algorithm. This marking process involves setting up a curve fitting for the two-dimensional lines. This curve fitting is performed using the least squares method, and at the same time, a preliminary judgment is made on the mineral through geometric judgment. This allows the determination of whether the edge curve on the ore is calculated using an analytical formula, which facilitates functional expression.

[0045] In this process, since some ores have relatively complete shapes and structures and their edge curves conform to relevant curve equations, the identification algorithm also uses ore model data for detection. The ore model data is a big data model that stores the basic characteristics of various existing ores, including basic characteristics that can be judged by the naked eye, such as shape and color. Setting up this big data model can accurately determine the type of ore and foreign objects. The judgment result is located by the identification algorithm, which facilitates the grasping device to grasp it.

[0046] The above-mentioned ore database model is set up. This database model, combined with the recognition algorithm, can better identify and compare with the content of the first image, classify and identify minerals, and understand the distribution of ores and foreign objects in the minerals. When a mineral matches the ore model data, the mineral is marked as an ore; when a mineral does not match the ore model data, the mineral is marked as a foreign object.

[0047] Since this application contains many foreign objects, these foreign objects are usually mixed with large iron pieces such as iron plates, steel bars, steel fibers, and shovel teeth. These foreign objects usually have very regular shapes. With the support of the recognition algorithm, the edges of such foreign objects can be fitted with two-dimensional curves. The result of this fitting is usually expressed by a function expression. When it conforms to the function expression, the recognition algorithm can mark the foreign object according to the function expression, so that the foreign object is separated from the ore. When the foreign object marking and the ore marking are completed, the content in the image is the content of the second image. At the same time, to ensure accuracy, the marked foreign object is confirmed by two-dimensional curve fitting from multiple angles in a three-dimensional rectangular coordinate system. When it is confirmed to be a foreign object, the grasping device grasps it according to the foreign object marking.

[0048] The second image content mentioned above includes the coordinate values ​​of the marked ore and the coordinate values ​​and function values ​​of the marked foreign object in multi-angle states, as well as the image content of the foreign object mark and the ore mark.

[0049] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for detecting and identifying foreign objects in ores, characterized in that, include: The camera assembly uses structured light to capture images of the mineral, collecting three-dimensional structural images of the mineral. The camera assembly includes an industrial camera and a projector, both of which are mounted on the gripping device. The projector emits light of a specific wavelength to illuminate the mineral, and the light reflected from the mineral is captured by the industrial camera to generate the three-dimensional structural image. The three-dimensional structural image is transmitted to the control terminal. The recognition algorithm performs three-dimensional image capture on the three-dimensional structural image, marks the ores and foreign objects in the three-dimensional structural image, and generates a marked image. The marked image includes the image features of each ore and foreign object. The recognition algorithm locates and identifies the marked image and generates the first image content; The recognition algorithm performs three-dimensional recognition based on the image features in the first image content, and collects geometric features in the image. The recognition algorithm marks foreign objects based on the geometric features and generates second image content. The grasping device grasps the foreign object based on the content of the second image; The recognition algorithm generates a three-dimensional coordinate system from the marked image, and each of the ore and foreign objects in the marked image is located by coordinates. The identification algorithm obtains the location coordinates of each of the ores and the foreign objects, and generates the first image content; The identification algorithm for obtaining the coordinates of the ore and the foreign object includes: The identification algorithm scans images of the ore and the foreign object using the camera component; The recognition algorithm performs line fitting on the edge positions of the ore and the foreign object to generate two-dimensional lines; The recognition algorithm marks the two-dimensional lines with two-dimensional coordinates. The recognition algorithm expresses the two-dimensional lines with the two-dimensional coordinate points using a functional algorithm and then makes a judgment: When the two-dimensional line cannot be functionalized by the functionalization algorithm, the recognition algorithm marks the mineral corresponding to the two-dimensional line as ore; when the two-dimensional line is functionalized by the functionalization algorithm, the recognition algorithm marks the mineral corresponding to the two-dimensional line as foreign object.

2. The method for detecting and identifying foreign objects in ores according to claim 1, characterized in that, The recognition algorithm acquires the mineral's coordinates from multiple angles and fuses the acquired images using coordinate functions to determine geometric features. Based on these geometric features, the recognition algorithm generates a foreign object marker to generate the second image content.

3. The method for detecting and identifying foreign objects in ores according to claim 2, characterized in that, The second image content includes the coordinate values ​​of the marked ore and the coordinate values ​​and function values ​​of the marked foreign object.

4. The method for detecting and identifying foreign objects in ores according to claim 1, characterized in that, The identification method also includes ore model data. The identification algorithm performs big data comparison and identification on the content of the first image based on the ore model data. When the mineral matches the ore model data, the mineral is marked as the ore; when the mineral does not match the ore model data, the mineral is marked as the foreign object. The grasping device grasps the foreign object according to the foreign object marker.

5. A foreign object detection and identification device for ores, characterized in that, The identification device is used to perform the ore foreign object detection and identification method as described in any one of claims 1-4.

6. The ore foreign object detection and identification device according to claim 5, characterized in that, The identification device includes a gripping device, which includes an industrial camera and a projector.

Images