A quartz sand impurity rapid detection system and detection method

Abstract

The application relates to the technical field of detection systems, in particular to a quartz sand impurity rapid detection system, which comprises an image processing unit, which regularly shoots images on a conveying belt and records the images as flow images, and determines a target area corresponding to each flow image based on overlapping parts; an impurity identification unit, which identifies a material contour in the target area, analyzes an impurity type corresponding to the material contour, records the material contour corresponding to the impurity as a target contour, and determines the position of the target contour; and a jet control unit, which analyzes and determines a jet timing of each jet nozzle based on position coordinates when the target contour is identified and labeled, in combination with a conveying speed of the conveying belt; while outputting the jet timing corresponding to the target contour, the jet control unit analyzes shape features of the target contour to obtain a projection area and a thickness measurement value, and adjusts airflow parameters of the jet nozzle in combination with the corresponding related impurity type. The application can effectively distinguish different types of impurities and perform targeted rejection.

Description

Technical Field

[0001] This invention relates to the field of detection system technology, specifically to a rapid detection system and method for impurities in quartz sand. Background Technology

[0002] Quartz sand is quartz particles produced by crushing and screening quartz stone. Quartz stone is a non-metallic mineral, primarily composed of silicon dioxide. Quartz sand plays an indispensable role in several key industrial sectors. As a core raw material in the glass industry, quartz sand ensures the transparency and durability of glass products; in the ceramics and refractory materials fields, it enhances the high-temperature resistance and chemical stability of products; in the metallurgical industry, quartz sand is used as an additive to optimize alloy properties; and it is also an important raw material in many other industries such as construction and chemicals.

[0003] In existing technologies, when detecting, identifying, and removing impurities in quartz sand, an air-jet removal system based on two-dimensional image analysis is typically used. This system identifies impurities through image recognition technology and removes them by air jets. However, this system can only simply identify impurities that are not quartz sand and cannot effectively distinguish impurities with vastly different physical properties, resulting in the inability to perform targeted removal and a problem of poor efficiency. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides a rapid detection system and method for quartz sand impurities, which can effectively solve the problem that existing technologies cannot effectively distinguish different types of impurities and remove them in a targeted manner.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] This invention provides a rapid detection system for impurities in quartz sand, comprising at least:

[0007] The image processing unit sets the shooting cycle based on the conveyor belt's conveying speed, and takes images on the conveyor belt at shooting intervals based on the shooting cycle, recording them as logistics images. It acquires two adjacent logistics images and records them as adjacent image groups. It analyzes the overlapping part of two logistics images in the adjacent image group to determine whether the conveyor belt has shifted, and determines the target area corresponding to each logistics image based on the overlapping part.

[0008] The impurity identification unit identifies the material outline in the target area based on image recognition technology and determines the position coordinates of the material outline. It analyzes the image features corresponding to the material outline and constructs a visual feature set. It compares the visual feature set with multiple preset impurity feature sets to determine the impurity type corresponding to the material outline. The material outline corresponding to the impurity is recorded as the target outline.

[0009] The jet control unit analyzes and determines the jet timing of each jet nozzle based on the position coordinates when the target contour is identified and marked, combined with the conveyor belt speed.

[0010] While outputting the target profile corresponding to the jet timing, the shape characteristics of the target profile are analyzed to obtain the projected area and thickness measurement values. Combined with the analysis of the corresponding relevant impurity types, the influencing parameters are calculated. Based on the influencing parameters, the airflow parameters of the jet nozzle are adjusted at the corresponding jet timing.

[0011] Furthermore, the shooting schedule setting process is as follows:

[0012] The conveyor belt speed is obtained, and the length of the conveyor belt within the shooting angle of the high-speed industrial camera is recorded as the shooting length. A redundancy ratio is preset. The effective ratio is obtained by subtracting the redundancy ratio from 1. The shooting length is multiplied by the effective ratio to obtain the effective length. The shooting cycle is obtained by dividing the effective length by the conveyor belt speed.

[0013] Furthermore, the analysis process for the overlapping parts is as follows:

[0014] Two logistics images in adjacent image groups are labeled as the first image and the second image respectively according to the shooting order. Matching feature points of the first image and the second image are identified based on the feature detection algorithm. The second image is transformed to the coordinate system of the first image according to the position of the matching feature points. The proportion of the overlapping area of ​​the two images is recorded as the overlap ratio.

[0015] When the overlap ratio equals the redundancy ratio, the non-overlapping part in the second image is recorded as the target region.

[0016] When the overlap ratio is not equal to the redundancy ratio, a conveyor belt slip signal is generated, triggering a shutdown maintenance command and sending it to the control console.

[0017] Furthermore, the material contour recognition and analysis process is as follows:

[0018] In the second image, a Cartesian coordinate system is constructed and denoted as the reference coordinate system. The horizontal axis of the reference coordinate system is parallel to the width direction of the conveyor belt, and the vertical axis of the reference coordinate system coincides with the edge of the conveyor belt. The distance between the horizontal axis of the reference coordinate system and the end of the conveyor belt is recorded as the reference distance.

[0019] Based on image recognition technology, multiple material contours within the target area are identified. The images within each material contour are independently identified, the target contours are filtered out, and the center coordinates and impurity types corresponding to each target contour are obtained.

[0020] Furthermore, the feature set comparison process is as follows:

[0021] The image within the outline of the material is captured and converted to grayscale. The average grayscale value of the pixels within the outline of the material is recorded as the color feature value.

[0022] The longest line segment with both ends on the material outline is called the major axis. The line segment perpendicular to the major axis with both ends on the material outline is called the second major axis. The longest second major axis is selected and called the minor axis. The ratio of the major axis to the minor axis is calculated and called the proportional characteristic value.

[0023] The first shape feature value is obtained by multiplying the area of ​​the material outline by 4π and dividing it by the square of the perimeter of the material outline. The second shape feature value is obtained by dividing the area of ​​the material outline by the area of ​​its smallest bounding rectangle.

[0024] The standard deviation of the gray value of each pixel within the material outline is recorded as the rough feature value. The gradient amplitude of each pixel within the material outline is calculated and the variance of the gradient amplitude within the material outline is recorded as the fluctuation feature value.

[0025] The visual feature set of the material profile is composed of multiple feature values. There are multiple influencing impurities. Each influencing impurity is assigned a corresponding impurity feature set. The visual feature set of the material profile is compared with the impurity feature set one by one to calculate the similarity. There is a preset similarity threshold. When the similarity is greater than the similarity threshold, the corresponding influencing impurity is recorded as the relevant impurity of the material profile.

[0026] Furthermore, the similarity calculation process is as follows:

[0027] There are multiple preset individual evaluation values, each corresponding to a feature value. A difference threshold is set for each feature value. The difference between the values ​​of the same feature value in two feature sets is calculated and the absolute value is obtained. The absolute value is compared with the corresponding difference threshold. When the difference is greater than the difference threshold, the corresponding individual evaluation value is assigned a value of 0. When the difference is less than or equal to the difference threshold, the corresponding individual evaluation value is assigned a value of 1. The similarity is obtained by calculating the weighted sum of multiple individual evaluation values. The weight coefficients of all individual evaluation values ​​are preset values.

[0028] Furthermore, the process for determining the timing of jet injection is as follows:

[0029] Obtain the control range of each jet nozzle, assign a control range to each jet nozzle based on the interval range corresponding to the horizontal axis of the reference coordinate system, and obtain the center coordinates corresponding to the target contour.

[0030] Based on the control range to which the horizontal coordinate of the center coordinate belongs, the target contour is bound to the corresponding jet nozzle, the vertical coordinate of the center coordinate is obtained and multiplied by a preset distance coefficient to obtain the relative distance, and the relative distance is added to the reference distance to obtain the target distance;

[0031] The control time difference is obtained by dividing the target distance by the conveyor belt speed. The control time difference is added to the current time to obtain the control time. A jet timing set is constructed for each jet nozzle. The control time of the target contour is recorded into the corresponding jet timing set. The jet nozzle is jet controlled based on the jet timing set.

[0032] Furthermore, the process of adjusting the airflow parameters is as follows:

[0033] The area of ​​the target contour is recorded as the projected area. The density of multiple related impurities corresponding to the target contour is obtained and the maximum value among them is recorded as the target density.

[0034] Control the flash to illuminate the material and collect images to analyze shadows, and determine the thickness measurement value corresponding to each target contour;

[0035] The product of the target density, projected area, and thickness is calculated to obtain the measured mass corresponding to the target contour. The measured mass is then multiplied by a pressure adjustment coefficient to obtain the corresponding pressure parameter.

[0036] Furthermore, the process for obtaining the thickness measurement value corresponding to the target contour is as follows:

[0037] S1: After determining the target contour position coordinates, control the flash lamp set above the conveyor belt and illuminating downwards to flash once, and at the same time take a picture of the material on the conveyor belt and record it as the flash image. Based on the shooting interval between the flash image and the second image and the position of the target contour in the second image, determine the contour area corresponding to each target contour in the flash image and record it as the solid area.

[0038] S2: After grayscale processing of the flash image, obtain the projection area corresponding to each entity region, and record the part of the projection area other than the entity region as the ghost image region of the target contour.

[0039] S3: Draw the ray projection corresponding to the flash light path in the flash image, denoted as the light source ray. Obtain the farthest point of the shadow area in the direction of the light source ray, denoted as the first measurement point. With a preset shadow grayscale range, construct a shadow line segment from the first measurement point towards the target contour. Obtain the length of the shadow line segment, denoted as the shadow length. The shadow line segment satisfies the following conditions:

[0040] Condition 1: The grayscale values ​​of all pixels on the line segment are within the shaded grayscale range;

[0041] Condition 2: The shaded segment is the longest segment that satisfies Condition 1;

[0042] S4: Multiply the shadow length by a preset thickness conversion factor to obtain the thickness measurement value of the impurity corresponding to the target contour.

[0043] A rapid detection method for impurities in quartz sand includes the following steps:

[0044] Step 1: Set the shooting cycle based on the conveyor belt speed, analyze the overlapping part of two adjacent logistics images, and determine whether the conveyor belt has shifted.

[0045] The target area corresponding to each logistics image is determined based on the overlapping parts;

[0046] Step 2: Identify the material outline in the target area and determine the position of the material outline. Analyze the image features corresponding to the material outline and construct a visual feature set. Compare the visual feature set with multiple preset impurity feature sets to determine the impurity type corresponding to the material outline. Record the material outline corresponding to the impurity as the target outline.

[0047] Step 3: Based on the position coordinates of the target contour when it is identified and marked, combined with the conveyor belt speed, analyze and determine the timing of each jet nozzle's jet injection.

[0048] Step 4: While outputting the jet timing corresponding to the target contour, analyze the shape characteristics of the target contour to obtain the projected area and thickness measurement values. Combine the analysis and calculation of the corresponding relevant impurity types to obtain the influencing parameters. Based on the influencing parameters, adjust the airflow parameters of the jet nozzle at the corresponding jet timing.

[0049] The technical solution provided by this invention has the following advantages compared with the known prior art:

[0050] This invention, on the one hand, can perform similarity comparison calculations between the apparent images of material particles and the apparent features of various types of impurities, realizing quantitative evaluation in the impurity detection and screening process of materials. Thus, it outputs analysis results with data that can be directly recognized by machines, which can be used to determine the correlation between the two, thereby realizing the identification and screening of impurities. On the other hand, it can simulate the thickness of impurities, thereby combining the projected area of ​​impurities to estimate the mass of impurities and make targeted adjustments to the jetting data, so that the jetting nozzle can more effectively remove impurities, especially different types of impurities, solving the problem of difficulty in screening quartz ore containing multiple impurities in the prior art. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0052] Figure 1 This is an overall module block diagram of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] The present invention will be further described below with reference to embodiments.

[0055] See Figure 1 A rapid detection system for impurities in quartz sand is used to separate impurities in quartz sand and detect the impurity content. It includes at least a vibrating feeder, a flat belt conveyor, a high-speed industrial camera, and an air jet nozzle.

[0056] It should be noted that the vibrating feeder is located at the feed end of the conveyor belt, uniformly feeding the quartz sand material onto the conveyor belt and ensuring that the quartz sand material is distributed in a single layer on the conveyor belt. Multiple air jet nozzles are located directly above the end of the conveyor belt, arranged in a linear array along the width of the belt. Each nozzle is responsible for a specific narrow belt area and is controlled by an independent solenoid valve. When the quartz sand material falls at the end of the conveyor belt, air is blown through the air jet nozzles to blow impurities into the waste bin. The system also includes:

[0057] The image processing unit sets the shooting cycle based on the conveyor belt's conveying speed. It takes one image of the conveyor belt at each shooting cycle interval and records it as a logistics image. It acquires two adjacent logistics images (at the shooting time) and records them as adjacent image groups. It analyzes the overlapping part of the two logistics images in the adjacent image group to determine whether the conveyor belt has shifted. Based on the overlapping part, it performs image stitching processing on the images in the adjacent image group.

[0058] Specifically, the shooting schedule is set as follows:

[0059] The conveyor belt speed (i.e., the linear velocity of the material moving on the conveyor belt) is obtained. The length of the conveyor belt within the shooting angle of the high-speed industrial camera is recorded as the shooting length. A redundancy ratio is preset (20% in a specific embodiment). The effective ratio is obtained by subtracting the redundancy ratio from 1. The effective length is obtained by multiplying the shooting length by the effective ratio. The shooting cycle is obtained by dividing the effective length by the conveyor belt speed.

[0060] It should be noted that setting a redundancy ratio aims to retain the key parts of the logistics images, thereby ensuring that there is at least a certain overlap between two adjacent logistics images. This allows for the stitching and combination of images in adjacent image groups based on the overlap, which helps to accurately determine the location of each impurity. (An important issue that cannot be ignored when conveyor belts transport logistics is conveyor belt slippage. When there is insufficient tension, load changes, or changes in the friction coefficient of the drive rollers, the conveyor belt may slip, causing misalignment between adjacent images. If a redundancy ratio is not set for stitching, there will be missed images, affecting the triggering timing of the air jet nozzle.)

[0061] Specifically, the analysis process for the overlapping parts is as follows:

[0062] Two logistics images in adjacent image groups are labeled as the first image and the second image respectively according to the shooting order. Matching feature points of the first image and the second image are identified based on feature detection algorithms (such as SIFT, SURF and ORB). The second image is transformed to the coordinate system of the first image according to the position of the matching feature points (ensuring that the corresponding matching feature points in the two images are in the same position). The proportion of the overlapping area of ​​the two images is recorded as the overlap ratio.

[0063] When the overlap ratio equals the redundancy ratio, the non-overlapping part in the second image is recorded as the target area, and further identification and analysis are performed on the materials within the target area.

[0064] When the overlap ratio is not equal to the redundancy ratio, a conveyor belt slip signal is generated, triggering a shutdown and maintenance command which is sent to the control console. Staff then perform shutdown and maintenance to ensure stable transport of the conveyor belt.

[0065] It should be noted that the feature point recognition, feature point matching, position change, and image fusion techniques used in the image stitching process are existing technologies and will not be elaborated on further here.

[0066] By comparing the overlap ratio and the redundancy ratio, it is possible to identify whether there is relative displacement (i.e., slippage) between the conveyor belt and the drive roller. This triggers a maintenance command to guide personnel to perform maintenance, preventing the conveyor belt slippage from affecting impurity identification and screening, and ensuring the accuracy of the air blowing time of the jet nozzle. It is worth noting that when the overlap ratio and the redundancy ratio are always equal, the conveyor belt should be considered to be operating stably, and the quartz sand material carried on it is moving at a stable linear velocity.

[0067] The impurity identification unit uses image recognition technology to identify and analyze the material contours within the target area in adjacent image groups, and determines the image information of multiple impurities and their position information (relative to the industrial camera).

[0068] Specifically, the material contour recognition and analysis process is as follows:

[0069] In the second image, a Cartesian coordinate system is constructed and denoted as the reference coordinate system. The horizontal axis of the reference coordinate system is parallel to the width direction of the conveyor belt, and the vertical axis of the reference coordinate system coincides with the edge of the conveyor belt (on one side). The distance between the horizontal axis of the reference coordinate system and the end of the conveyor belt is recorded as the reference distance (the distance from any point in the second image to the end of the conveyor belt can be determined by the reference distance). Based on image recognition technology, multiple material contours in the target area are identified (each material contour corresponds to an independent one on the conveyor belt). The image within each material contour is independently identified (to determine whether the material contour belongs to an impurity). The material contours corresponding to the impurities are selected and recorded as target contours. The center coordinates and impurity type corresponding to each target contour are obtained.

[0070] It should be noted that there is a significant color contrast between the conveyor belt and the quartz sand material. Therefore, image recognition algorithms can quickly identify the scattered material on the conveyor belt and outline the contours of individual materials. It is important to note that this quartz sand impurity detection system is only suitable for quartz mineral materials that can be clearly captured and identified by an optical camera; it cannot accurately identify quartz sand with fine particles and large quantities.

[0071] More specifically, the independent identification process for material outlines is as follows:

[0072] The image within the outline of the material is captured and converted to grayscale. The average grayscale value of the pixels within the outline of the material is recorded as the color feature value.

[0073] The longest line segment with both ends on the material outline is called the major axis. The line segment perpendicular to the major axis with both ends on the material outline is called the second major axis. The longest second major axis is selected and called the minor axis. The ratio of the major axis to the minor axis is calculated and called the proportional characteristic value.

[0074] Multiply the area of ​​the material outline by 4π and divide by the square of the perimeter of the material outline to obtain the first shape feature value (when the first shape feature value is approximately equal to 1, it means that the material outline is close to a circle). Divide the area of ​​the material outline by the area of ​​its smallest bounding rectangle (i.e. the smallest rectangle covering the range of the material outline) to obtain the second shape feature value (when the second shape feature value is approximately equal to 1, it means that the material outline is close to a rectangle).

[0075] The standard deviation of the gray values ​​of each pixel within the material outline is recorded as the roughness feature value (the larger the value, the rougher the material surface). The gradient magnitude of each pixel within the material outline is calculated, and the variance of the gradient magnitude within the material outline is recorded as the undulation feature value (this value reflects the degree of undulation of the material surface; the larger the gradient variance, the greater the degree of undulation).

[0076] The visual feature set of the material contour is composed of multiple feature values ​​(color feature value, proportion feature value, first shape feature value, second shape feature value, roughness feature value, and undulation feature value). Multiple influencing impurities are preset, and each influencing impurity has a corresponding impurity feature set (also containing color feature value, proportion feature value, first shape feature value, second shape feature value, roughness feature value, and undulation feature value; specific values ​​are set by staff based on experience). The visual feature set of the material contour is compared one by one with the multiple impurity feature sets to calculate the similarity between the two feature sets. A similarity threshold is preset. When the similarity is greater than the similarity threshold, the corresponding influencing impurity is recorded as a related impurity of the material contour (the same material contour may correspond to multiple related impurities). When a material contour has related impurities, it is recorded as the target contour (i.e., the target contour screening process).

[0077] It should be noted that when the similarity between two feature sets is greater than the similarity threshold, it means that there are obvious similar image features between the corresponding material outline and the influencing impurities. These features include, but are not limited to, color, shape, texture, etc., thereby filtering out multiple related impurities corresponding to the material outline, that is, the possible material types of the material outline. This step can use the appearance features shown by the material outline to compare with the appearance features of known impurities, thereby determining the type of impurity corresponding to the material outline, so that the impurity can be removed by blowing air in the subsequent process, realizing the sorting and screening of materials.

[0078] It should be noted that the gradient magnitude calculation process first uses the Sobel operator to calculate the directional gradient (X-direction gradient Gx) and vertical directional gradient (Y-direction gradient Gy) of each pixel, and then uses the synthesis formula (G=√(Gx²+Gy²)) to calculate the total gradient magnitude as the gradient magnitude of that pixel.

[0079] Furthermore, the similarity calculation process is as follows:

[0080] There are multiple preset individual evaluation values, each corresponding to a feature value. A difference threshold is set for each feature value. The difference between the values ​​of the same feature value in two feature sets is calculated and the absolute value is obtained. The absolute value is compared with the corresponding difference threshold. When the difference is greater than the difference threshold, the corresponding individual evaluation value is assigned a value of 0. When the difference is less than or equal to the difference threshold, the corresponding individual evaluation value is assigned a value of 1. The similarity is obtained by calculating the weighted sum of multiple individual evaluation values. The weight coefficients of all individual evaluation values ​​are preset values ​​(set by staff based on experience).

[0081] Specifically, let the two feature sets be respectively , The corresponding individual evaluation value is The corresponding difference threshold is The corresponding weighting coefficient is The formula for assigning a single evaluation value is expressed as follows: If n = 1, 2, ..., 6, then the formula for calculating similarity is expressed as: .

[0082] It should be noted that by calculating similarity, the data similarity of elements in two feature sets can be quantitatively evaluated, so as to output analysis results with data that can be directly recognized by machines, which can be used to judge the correlation between the two, and thus realize the identification and screening of impurities.

[0083] The jet control unit analyzes and determines the jet timing of each jet nozzle based on the position coordinates of the target contour when it is identified and marked, combined with the conveyor belt speed. While outputting the jet timing corresponding to the target contour, it combines the shape characteristics of the target contour with the corresponding impurity types and analyzes them to obtain the influencing parameters. Based on the influencing parameters, it adjusts the airflow parameters of the jet nozzle.

[0084] Specifically, the process for determining the timing of jet injection is as follows:

[0085] Obtain the control range of each jet nozzle. Based on the interval range corresponding to the control range of the jet nozzle on the horizontal axis of the reference coordinate system, assign a control interval to each jet nozzle. Obtain the center coordinates corresponding to the target contour. Based on the control interval to which the horizontal coordinate of the center coordinate belongs, bind the target contour to the corresponding jet nozzle (i.e., the impurities corresponding to the target contour are removed by the jet nozzle it is bound to). Obtain the vertical coordinate of the center coordinate and multiply it by a preset distance coefficient (i.e., the ratio of the actual distance to the unit length in the coordinate system) to obtain the relative distance of the target contour from the horizontal axis of the reference coordinate system (this distance value may be negative). Add the relative distance to the reference distance to obtain the target distance. Divide the target distance by the conveyor belt speed to obtain the control time difference. Add the control time difference to the current time to obtain the control time. Construct a jet timing set for each jet nozzle. Record the control time of the target contour into the corresponding jet timing set. Perform jet control on the jet nozzle based on the jet timing set.

[0086] It should be noted that by constructing a set of jet timings for each jet nozzle, independent control of a single jet nozzle can be achieved (specific control relies on a solenoid valve). This allows the jet nozzle to jet in an orderly manner according to the time nodes within the set of jet timings, sequentially blowing impurities passing diagonally below the jet nozzle into the waste bin for collection, thereby improving the purity of the quartz sand obtained from screening.

[0087] Specifically, the process for adjusting the airflow parameters is as follows:

[0088] The area of ​​the target contour is recorded as the projected area. The density of multiple related impurities corresponding to the target contour is obtained and the maximum value is selected as the target density (the airflow parameters obtained by substituting the maximum value as the target density can be applied to all possible related impurities). The flash lamp is controlled to irradiate the material and the image is collected and the shadow is analyzed to determine the thickness measurement value corresponding to each target contour. The product of the target density, projected area and thickness measurement value is calculated to obtain the measured mass corresponding to the target contour. The measured mass is multiplied by a pressure adjustment coefficient to obtain the corresponding pressure parameter (in this calculation process, the measured mass is a dimensionless value and the pressure adjustment coefficient is in the unit of pressure). The air pressure of the jet nozzle during the corresponding control time is set to be equal to the pressure parameter.

[0089] By introducing thickness measurement and density estimation, impurities are analyzed and evaluated from a three-dimensional perspective. Compared with the two-dimensional analysis perspective in existing technologies, this can more accurately reflect the actual data of impurities, especially the mass, which is difficult to measure. Based on this, the air pressure is calculated to ensure that different types of impurities can be effectively removed. In particular, for impurities with large mass but small projected area, sufficient airflow can be generated to remove them, significantly improving the removal rate.

[0090] More specifically, the process for obtaining the thickness measurement value corresponding to the target contour is as follows:

[0091] After analyzing the second image to determine the target contour's position coordinates, a flash lamp positioned above the conveyor belt and pointing downwards is immediately activated, flashing once. Simultaneously, an image of the material on the conveyor belt is captured and recorded as the flash image. Based on the shooting interval between the flash image and the second image, combined with the position of the target contour in the second image, the contour region corresponding to each target contour in the flash image is determined and recorded as the solid region (this region is obtained by translating the target contour; the translation distance is equal to the conveyor speed multiplied by the shooting interval). After grayscale processing of the flash image, the projection region corresponding to each solid region is obtained (the projection region is the horizontal projection region of the impurity and its shadow corresponding to the target contour). The area in the projection region excluding the solid region is recorded as the target. The shadow region of the contour (the area outside the solid region of the projection area is the shadow region of the impurity) is plotted in the flash image as the ray projection corresponding to the flash lamp path, denoted as the light source ray. The farthest point of the shadow region in the direction of the light source ray is obtained and denoted as the first measurement point. A preset shadow grayscale range (representing the range of grayscale value changes of the shadow part in the grayscale image, and pixels within this range are identified as shadow pixels) is constructed from the first measurement point towards the target contour direction, with a shadow line segment (one pixel wide). The length of the shadow line segment is obtained and denoted as the shadow length. The shadow length is multiplied by a preset thickness conversion factor to obtain the thickness measurement value of the impurity corresponding to the target contour. The shadow line segment satisfies the following conditions:

[0092] Condition 1: The grayscale values ​​of all pixels on the line segment are within the shaded grayscale range;

[0093] Condition 2: The shaded line segment is the longest line segment that satisfies Condition 1.

[0094] It should be noted that the shaded line segment usually corresponds to the line connecting the highest point of the impurity and the farthest point of the shadow. Therefore, by determining the length of the shaded line segment and combining it with the illumination angle of the light source, the height of the highest point of the impurity can be calculated using trigonometric functions in geometry. Based on this height, the thickness of the impurity can be simulated, which helps to analyze and estimate the quality of the impurity. This allows for targeted adjustments to the jetting data, enabling the jetting nozzle to more effectively remove impurities, especially different types of impurities. This solves the problem in existing technologies of difficulty in screening quartz ore containing multiple impurities.

[0095] A rapid detection system and method for impurities in quartz sand includes the following steps:

[0096] Step 1: Set the shooting cycle based on the conveyor belt speed, and use the shooting cycle as the shooting interval to shoot images on the conveyor belt and record them as logistics images. Acquire two adjacent logistics images and record them as adjacent image groups. Analyze the overlapping part of the two logistics images in the adjacent image group to determine whether the conveyor belt has shifted, and determine the target area corresponding to each logistics image based on the overlapping part.

[0097] By analyzing the overlapping parts of adjacent images, the conveyor belt can be slipped, triggering maintenance commands, thereby ensuring the accuracy of the target area's coordinate position and improving the accuracy of impurity positioning.

[0098] Step 2: Identify the material outline in the target area based on image recognition technology and determine the position of the material outline. Analyze the image features corresponding to the material outline and construct a visual feature set. Compare the visual feature set with multiple preset impurity feature sets to determine the impurity type corresponding to the material outline. Record the material outline corresponding to the impurity as the target outline.

[0099] By constructing a set of visual features to reflect the image characteristics of materials and comparing them with the features of various preset impurities, the accuracy of impurity type identification is improved.

[0100] Step 3: Based on the position coordinates of the target contour when it is identified and marked, combined with the conveyor belt speed, analyze and determine the timing of each jet nozzle's jet injection.

[0101] Step 4: While outputting the jet timing corresponding to the target contour, analyze the shape characteristics of the target contour to obtain the projected area and thickness measurement values. Combine the analysis and calculation of the corresponding relevant impurity types to obtain the influencing parameters. Based on the influencing parameters, adjust the airflow parameters of the jet nozzle at the corresponding jet timing.

[0102] Existing technologies often use fixed airflow parameters, which cannot cope with impurities of different characteristics. This method can dynamically adjust the airflow according to the projected area and thickness of the impurities, accurately matching the impurities with the removal intensity. This not only improves the removal efficiency but also avoids energy waste and material splashing caused by excessive jetting, achieving intelligent, adaptive, and efficient sorting.

[0103] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A rapid detection system for impurities in quartz sand, characterized in that, include: The image processing unit sets the shooting cycle based on the conveyor belt's conveying speed, and takes images on the conveyor belt at shooting intervals based on the shooting cycle, recording them as logistics images. It acquires two adjacent logistics images and records them as adjacent image groups. It analyzes the overlapping part of two logistics images in the adjacent image group to determine whether the conveyor belt has shifted, and determines the target area corresponding to each logistics image based on the overlapping part. The impurity identification unit identifies the material outline in the target area based on image recognition technology and determines the position coordinates of the material outline. It analyzes the image features corresponding to the material outline and constructs a visual feature set. It compares the visual feature set with multiple preset impurity feature sets to determine the impurity type corresponding to the material outline. The material outline corresponding to the impurity is recorded as the target outline. The jet control unit analyzes and determines the jet timing of each jet nozzle based on the position coordinates when the target contour is identified and marked, combined with the conveyor belt speed. While outputting the target profile corresponding to the jet timing, the shape characteristics of the target profile are analyzed to obtain the projected area and thickness measurement values. Combined with the analysis and calculation of the corresponding relevant impurity types, the influencing parameters are obtained. Based on the influencing parameters, the airflow parameters of the jet nozzle are adjusted at the corresponding jet timing. The process for adjusting airflow parameters is as follows: The area of ​​the target contour is recorded as the projected area. The density of multiple related impurities corresponding to the target contour is obtained and the maximum value among them is recorded as the target density. Control the flash to illuminate the material and collect images to analyze shadows, and determine the thickness measurement value corresponding to each target contour; The product of the target density, projected area, and thickness is calculated to obtain the measured mass corresponding to the target contour. The measured mass is then multiplied by a pressure adjustment coefficient to obtain the corresponding pressure parameter.

2. The rapid detection system for impurities in quartz sand according to claim 1, characterized in that, The shooting schedule setting process is as follows: The conveyor belt speed is obtained, and the length of the conveyor belt within the shooting angle of the high-speed industrial camera is recorded as the shooting length. A redundancy ratio is preset. The effective ratio is obtained by subtracting the redundancy ratio from 100%. The shooting length is multiplied by the effective ratio to obtain the effective length. The effective length is divided by the conveyor belt speed to obtain the shooting cycle.

3. The rapid detection system for impurities in quartz sand according to claim 1, characterized in that, The analysis process for the overlapping parts is as follows: Two logistics images in adjacent image groups are labeled as the first image and the second image respectively according to the shooting order. Matching feature points of the first image and the second image are identified based on the feature detection algorithm. The second image is transformed to the coordinate system of the first image according to the position of the matching feature points. The proportion of the overlapping area of ​​the two images is recorded as the overlap ratio. When the overlap ratio equals the redundancy ratio, the non-overlapping part in the second image is recorded as the target region. When the overlap ratio is not equal to the redundancy ratio, a conveyor belt slip signal is generated, triggering a shutdown maintenance command and sending it to the control console.

4. The rapid detection system for impurities in quartz sand according to claim 3, characterized in that, The material contour recognition and analysis process is as follows: In the second image, a Cartesian coordinate system is constructed and denoted as the reference coordinate system. The horizontal axis of the reference coordinate system is parallel to the width direction of the conveyor belt, and the vertical axis of the reference coordinate system coincides with the edge of the conveyor belt. The distance between the horizontal axis of the reference coordinate system and the end of the conveyor belt is recorded as the reference distance. Based on image recognition technology, multiple material contours within the target area are identified. The images within each material contour are independently identified, the target contours are filtered out, and the center coordinates and impurity types corresponding to each target contour are obtained.

5. The rapid detection system for impurities in quartz sand according to claim 4, characterized in that, The feature set comparison process is as follows: The image within the outline of the material is captured and converted to grayscale. The average grayscale value of the pixels within the outline of the material is recorded as the color feature value. The longest line segment with both ends on the material outline is called the major axis. The line segment perpendicular to the major axis with both ends on the material outline is called the second major axis. The longest second major axis is selected and called the minor axis. The ratio of the major axis to the minor axis is calculated and called the proportional characteristic value. The first shape feature value is obtained by multiplying the area of ​​the material outline by 4π and dividing it by the square of the perimeter of the material outline. The second shape feature value is obtained by dividing the area of ​​the material outline by the area of ​​its smallest bounding rectangle. The standard deviation of the gray value of each pixel within the material outline is recorded as the rough feature value. The gradient amplitude of each pixel within the material outline is calculated and the variance of the gradient amplitude within the material outline is recorded as the fluctuation feature value. The visual feature set of the material profile is composed of multiple feature values. There are multiple influencing impurities. Each influencing impurity is assigned a corresponding impurity feature set. The visual feature set of the material profile is compared with the impurity feature set one by one to calculate the similarity. There is a preset similarity threshold. When the similarity is greater than the similarity threshold, the corresponding influencing impurity is recorded as the relevant impurity of the material profile.

6. The rapid detection system for impurities in quartz sand according to claim 5, characterized in that, The similarity calculation process is as follows: There are multiple preset individual evaluation values, each corresponding to a feature value. A difference threshold is set for each feature value. The difference between the values ​​of the same feature value in two feature sets is calculated and the absolute value is obtained. The absolute value is compared with the corresponding difference threshold. When the difference is greater than the difference threshold, the corresponding individual evaluation value is assigned a value of 0. When the difference is less than or equal to the difference threshold, the corresponding individual evaluation value is assigned a value of 1. The similarity is obtained by calculating the weighted sum of multiple individual evaluation values. The weight coefficients of all individual evaluation values ​​are preset values.

7. The rapid detection system for impurities in quartz sand according to claim 6, characterized in that, The process for determining the timing of jet injection is as follows: Obtain the control range of each jet nozzle, assign a control range to each jet nozzle based on the interval range corresponding to the horizontal axis of the reference coordinate system, and obtain the center coordinates corresponding to the target contour. Based on the control range to which the horizontal coordinate of the center coordinate belongs, the target contour is bound to the corresponding jet nozzle, the vertical coordinate of the center coordinate is obtained and multiplied by a preset distance coefficient to obtain the relative distance, and the relative distance is added to the reference distance to obtain the target distance; The control time difference is obtained by dividing the target distance by the conveyor belt speed. The control time difference is added to the current time to obtain the control time. A jet timing set is constructed for each jet nozzle. The control time of the target contour is recorded into the corresponding jet timing set. The jet nozzle is jet controlled based on the jet timing set.

8. The rapid detection system for impurities in quartz sand according to claim 7, characterized in that, The process for obtaining the thickness measurement value corresponding to the target contour is as follows: S1: After determining the target contour position coordinates, control the flash lamp set above the conveyor belt and illuminating downwards to flash once, and at the same time take a picture of the material on the conveyor belt and record it as the flash image. Based on the shooting interval between the flash image and the second image and the position of the target contour in the second image, determine the contour area corresponding to each target contour in the flash image and record it as the solid area. S2: After grayscale processing of the flash image, obtain the projection area corresponding to each entity region, and record the part of the projection area other than the entity region as the ghost image region of the target contour. S3: Draw the ray projection corresponding to the flash light path in the flash image, denoted as the light source ray. Obtain the farthest point of the shadow area in the direction of the light source ray, denoted as the first measurement point. With a preset shadow grayscale range, construct a shadow line segment from the first measurement point towards the target contour. Obtain the length of the shadow line segment, denoted as the shadow length. The shadow line segment satisfies the following conditions: Condition 1: The grayscale values ​​of all pixels on the line segment are within the shaded grayscale range; Condition 2: The shaded segment is the longest segment that satisfies Condition 1; S4: Multiply the shadow length by a preset thickness conversion factor to obtain the thickness measurement value of the impurity corresponding to the target contour.

9. A rapid detection method for impurities in quartz sand, applied to the rapid detection system for impurities in quartz sand according to any one of claims 1-8, characterized in that, Includes the following steps: Step 1: Set the shooting cycle based on the conveyor belt speed, analyze the overlapping part of two adjacent logistics images, and determine whether the conveyor belt has shifted. The target area corresponding to each logistics image is determined based on the overlapping parts; Step 2: Identify the material outline in the target area and determine the position of the material outline. Analyze the image features corresponding to the material outline and construct a visual feature set. Compare the visual feature set with multiple preset impurity feature sets to determine the impurity type corresponding to the material outline. Record the material outline corresponding to the impurity as the target outline. Step 3: Based on the position coordinates of the target contour when it is identified and marked, combined with the conveyor belt speed, analyze and determine the timing of each jet nozzle's jet injection. Step 4: While outputting the jet timing corresponding to the target contour, analyze the shape characteristics of the target contour to obtain the projected area and thickness measurement values. Combine the analysis and calculation of the corresponding relevant impurity types to obtain the influencing parameters. Based on the influencing parameters, adjust the airflow parameters of the jet nozzle at the corresponding jet timing.

Images