CN110844518A - Conveying belt tearing detection method based on FPGA - Google Patents

Abstract

The invention discloses a conveying belt tearing detection method based on an FPGA (field programmable gate array), which comprises the following steps of: recording a real-time video when the conveyer belt runs, preprocessing images frame by frame in the video, sequentially sending each frame of image into a frame cache module, simultaneously extracting a standard image and an image to be detected from the frame cache module to be used as a background difference algorithm, and comparing a return value of the algorithm with a threshold value to realize the real-time detection of the tearing of the conveyer belt. The method realizes real-time detection of 'tearing' damage through the OV5640 image sensor and the FPGA chip, avoids the use of a high-performance computer and a large number of sensors, greatly saves cost, effectively improves identification precision by using the FPGA to perform a background difference algorithm, and reduces the error rate of identification.

Description

Conveying belt tearing detection method based on FPGA

Technical Field

The invention relates to a detection method for detecting the tearing of a conveyer belt by utilizing a background difference defect detection algorithm based on an FPGA (field programmable gate array).

Background

The belt conveyer is the main equipment for bulk material transportation and is widely used in the industrial fields of mines, chemical industry, grain production and the like. The conveying belt on the belt conveyor is inevitably subjected to tearing damage in the running process. If the damage is treated in time, many safety accidents can be avoided, but if the damage is not treated properly, the normal use of the conveying belt is seriously influenced, and even a series of safety accidents are caused. Some existing methods mainly utilize the computing power of a computer to compare a detection picture of a conveying belt with an original damage picture in a system, and not only a computer with strong computing power is needed, but also a large number of sensors are needed to be installed, so that the installation is complicated, the cost is high, and the effect is not ideal.

Disclosure of Invention

Aiming at the problems, the invention aims to provide a detection method for the 'tearing' damage of a conveyer belt based on an FPGA chip. When avoiding installing a large amount of sensor check out test set, realize the real-time supervision to the conveyer belt damage, reduced the running cost, can effectively avoid the incident moreover.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

the conveying belt tearing detection method based on the FPGA is characterized by comprising the following steps:

(1) a camera is arranged above the conveying belt to record images on the conveying belt uninterruptedly for a long time, brightness of the obtained images is judged, and output voltage of an LED light supplement lamp is adjusted, so that the recorded videos in different environments can be kept clear;

(2) through the programming of a Verilog HDL hardware description language, the logic circuit of the FPGA carries out real-time preprocessing on the received image data;

(2a) taking pictures from the recorded video of the conveying belt frame by frame;

(2b) performing gray-scale transformation on the obtained picture by using f (i, j) ═ 0.2989R (i, j) +0.5870G (i, j) +0.1140B (i, j), wherein f (i, j) represents a gray-scale value, R (i, j) represents a red component value, G (i, j) represents a green component value, and B (i, j) represents a blue component value;

(2c) performing median filtering on the image subjected to gray level transformation by using g (i, j) ═ mean { f (i-k, j-l) }, (k, l) ∈ H, wherein (i, j) represents a position coordinate of a certain pixel of the original image, g (i, j) represents a pixel value after gray level transformation on the position of the original image (i, j), f (i-k, j-l), (j, l) ∈ H represents each pixel value of a filtering template window H on the pixel of the original image (i, j), and mean { } represents a middle value of all pixel values in the selected { };

(2d) using | G | ═ G for median filtered pictures_x|+|G_y|，

Performing edge detection, wherein A represents the original image subjected to median filtering, G represents the image subjected to edge detection, and G_xRepresenting the gray value of the image subjected to horizontal edge detection, G_yRepresenting the gray value of the image detected by the vertical edge;

(2e) using P-P for picture subjected to edge detection₁₁&P₁₂&P₁₃&P₂₁&P₂₂&P₂₃&P₃₁&P₃₂&P₃₃Performing corrosion calculation, wherein P is the central pixel point of the corrosion calculation, P₁₁-P₃₃Corrosion calculation of 9 pixels in a central pixel 3 x 3 pixel matrix;

(3) connecting the FPGA chip with a DDR3 SDRAM, and using the DDR3 SDRAM chip as a frame buffer module for storing image data to be detected and standard image data;

(4) storing the standard images of the conveying belt recorded in advance under a good illumination condition into a frame cache module before starting detection;

(5) simultaneously extracting a standard image and an image to be detected from a frame buffer module, taking the standard image as a background frame, taking the image to be detected as a current image frame, and using the two images

Performing a background difference algorithm, wherein X_ijBeing pixel points of the current image frame, Y_ijThe theta is a return value obtained by a background difference algorithm;

(6) judging whether the return value theta obtained after subtraction is larger than a tearing threshold value, if so, judging that tearing occurs, and if not, waiting for the next frame of image to be detected and the standard image to carry out differential processing;

(7) if the volume is judged to be 'tearing', comparing theta with delta, marking by using a marking machine if delta is larger than theta, and if delta is smaller than or equal to theta, emergency stopping and giving an alarm, wherein delta 'tearing' is a volume threshold;

(8) and transmitting the picture judged to be torn to a display through the HDMI.

The invention has the beneficial effects that:

the conveying belt of the belt conveyor is often torn and damaged due to sharp appearance of conveyed articles, uneven weight distribution and the like in the using process. For such damage, if the damage cannot be timely found and repaired and replaced, the damage will cause great harm to the production safety. The method can monitor and discover the tearing damage in real time and provide timely and effective information for subsequent processing. The invention uses video information to detect and identify, the required hardware has small volume and convenient installation, and only one OV5640 image sensor and one FPGA core board are required. The traditional image detection equipment usually uses a high-performance computer to perform real-time data processing, and the FPGA is used as hardware equipment with a reconfigurable structure, so that the price is low, and the problem that the real-time performance is influenced because the data cannot be processed in time due to insufficient computing capacity of a chip is effectively solved because the FPGA is realized through a hardware circuit. The invention not only can conveniently detect the damage by using the background difference algorithm, but also can easily extend the defect detection under more conditions, and can realize the extension only by selecting a proper standard template image as a background frame according to the detected object. And the FPGA has lower cost, convenient installation and small volume, and can be put into production after being electrified after being successfully installed, thereby greatly improving the detection efficiency and saving the cost.

Drawings

FIG. 1 is a block diagram of the overall process of the method of the present invention.

FIG. 2 is a diagram of the hardware architecture of the system of the method of the present invention.

FIG. 3 is an algorithmic flow chart of the method of the present invention.

FIG. 4 is a flow chart of a background difference algorithm of the method of the present invention.

Detailed Description

As shown in fig. 1, the process of the conveying belt tearing detection method based on the FPGA is as follows:

(1) an OV5640 is arranged above the conveying belt to carry out long-term uninterrupted video recording on the conveying belt, and then the video is transmitted to an image acquisition module through an IIC bus;

(2) through the programming of a Verilog HDL hardware description language, the logic circuit of the FPGA carries out real-time preprocessing on the received image data;

(2a) taking pictures from the recorded video of the conveying belt frame by frame;

(2b) performing gray-scale transformation on the obtained picture by using f (i, j) ═ 0.2989R (i, j) +0.5870G (i, j) +0.1140B (i, j), wherein f (i, j) represents a gray-scale value, R (i, j) represents a red component value, G (i, j) represents a green component value, and B (i, j) represents a blue component value;

(2c) performing median filtering on the image subjected to gray level transformation by using g (i, j) ═ mean { f (i-k, j-l) }, (k, l) ∈ H, wherein (i, j) represents a position coordinate of a certain pixel of the original image, g (i, j) represents a pixel value after gray level transformation on the position of the original image (i, j), f (i-k, j-l), (j, l) ∈ H represents each pixel value of a filtering template window H on the pixel of the original image (i, j), and mean { } represents a middle value of all pixel values in the selected { };

(2d) using | G | ═ G for median filtered pictures_x|+|G_y|，

Performing edge detection, wherein A represents the original image subjected to median filtering, G represents the image subjected to edge detection, and G_xRepresenting the gray value of the image subjected to horizontal edge detection, G_yRepresenting the gray value of the image detected by the vertical edge;

(2e) using P-P for picture subjected to edge detection₁₁&P₁₂&P₁₃&P₂₁&P₂₂&P₂₃&P₃₁&P₃₂&P₃₃Performing corrosion calculation, wherein P is the central pixel point of the corrosion calculation, P₁₁-P₃₃Is calculation of corrosion9 pixels in a central pixel 3 × 3 pixel matrix;

(3) storing the preprocessed image data and the pre-recorded standard images of the conveyor belt into a frame cache module, wherein an FPGA chip is connected with a DDR3 SDRAM, and a DDR3 SDRAM chip is used as the frame cache module;

(4) simultaneously extracting a standard image and an image to be detected from a frame buffer module, taking the standard image as a background frame, taking the image to be detected as a current image frame, and using the two images

Performing a background difference algorithm, wherein X_ijBeing pixel points of the current image frame, Y_ijThe theta is a return value obtained by a background difference algorithm;

(5) judging whether the return value theta obtained after subtraction is larger than a tearing threshold value, if so, judging that tearing occurs, and if not, waiting for the next frame of image to be detected and the standard image to carry out differential processing;

(6) transmitting the picture which is judged to be torn to a display through an HDMI interface, comparing theta with delta, judging the size of the torn volume, marking by using a marking machine if delta is larger than theta, and emergently stopping and giving an alarm if delta is smaller than or equal to theta, wherein delta is used for tearing the volume threshold;

as shown in fig. 2, the system hardware structure of the method of the present invention is:

(1) judging the brightness of the acquired image, and adjusting the voltage of an LED light supplement lamp through an FPGA (field programmable gate array) to ensure that videos in different environments are sufficiently clear;

(2) the FPGA chip is connected with the OV5640 image sensor, and image information is transmitted to the FPGA chip through an IIC bus;

(3) connecting the FPGA chip with a DDR3 SDRAM, and using the DDR3 SDRAM chip as a frame buffer module for storing image data to be detected and standard image data;

(4) if the volume is judged to be 'tearing', comparing theta with delta, if delta is larger than theta, marking by using a marking machine, if delta is smaller than or equal to theta, carrying out emergency stop, and giving an alarm, wherein delta 'tearing' is a volume threshold, and theta is a background difference algorithm return value;

(5) the FPGA chip is connected with a display through an HDMI interface module and is used for displaying the torn picture;

as shown in fig. 3, the algorithm flow of the method of the present invention is:

(1) first using the acquired conveyor belt image

Carrying out gray scale transformation on the f (i, j) ═ 0.2989R (i, j) +0.5870G (i, j) +0.1140B (i, j), converting the collected original conveying belt RGB image into a gray scale image, and reducing the data processing amount, wherein f (i, j) represents a gray scale value, R (i, j) represents a red component value, G (i, j) represents a green component value, and B (i, j) represents a blue component value;

(2) carrying out median filtering on the conveyer belt gray level image by using g (i, j) ═ mean { f (i-k, j-l) }, (k, l) ∈ H, and eliminating noise interference in the image, wherein (i, j) represents a position coordinate of a certain pixel of the original picture, g (i, j) represents a pixel value after the position gray level of the original picture (i, j) is transformed, f (i-k, j-l), (j, l) ∈ H represents each pixel value of a filtering template window H on the pixel of the original picture (i, j), and mean { } represents a middle value of all pixel values in the selected { };

(3) then using G for the belt grayscale image_x|+|G_y|，

Performing edge detection to obtain a binary image of the conveyor belt, wherein A represents an original image subjected to median filtering, G represents an image subjected to edge detection, and G represents_xRepresenting the gray value of the image subjected to horizontal edge detection, G_yRepresenting the gray value of the image detected by the vertical edge;

(4) then using P ═ P for the obtained binary image₁₁&P₁₂&P₁₃&P₂₁&P₂₂&P₂₃&P₃₁&P₃₂&P₃₃Carrying out corrosion treatment to eliminate irrelevant detail parts and avoid the influence of slight difference on a conveyer belt on a detection result, wherein P is a central pixel point of corrosion calculation and P is a central pixel point of corrosion calculation₁₁-P₃₃Corrosion calculation of 9 pixels in a central pixel 3 x 3 pixel matrix;

(5) finally, use

Performing background difference algorithm, processing the binary image to be detected and the standard binary image to obtain the final algorithm return value, wherein X_ijBeing pixel points of the current image frame, Y_ijThe theta is a return value obtained by a background difference algorithm;

as shown in fig. 4, the calculation process of the background subtraction method used in the method of the present invention is as follows:

(1) respectively extracting pixel points of the standard template and pixel points of the image to be detected from the corresponding address area of the DDR3 SDRAM;

(2) taking the standard image as a background frame, taking the image to be detected as a current image frame, and using the two images

Performing background difference algorithm to compare each pixel point one by one, wherein X_ijBeing pixel points of the current image frame, Y_ijThe theta is a return value obtained by a background difference algorithm;

(3) and judging whether the return value theta obtained after subtraction is larger than a tearing threshold value, if so, judging that tearing occurs, and if not, waiting for the next frame of image to be detected and the standard image to carry out differential processing.

Claims (1)

1. The conveying belt tearing detection method based on the FPGA is characterized by comprising the following steps:

(1) a camera is arranged above the conveying belt to record images on the conveying belt uninterruptedly for a long time, brightness of the obtained images is judged, and output voltage of an LED light supplement lamp is adjusted, so that the recorded videos in different environments can be kept clear;

(2) through the programming of a Verilog HDL hardware description language, the logic circuit of the FPGA carries out real-time preprocessing on the received image data;

(2a) taking pictures from the recorded video of the conveying belt frame by frame;

(2b) performing gray-scale transformation on the obtained picture by using f (i, j) ═ 0.2989R (i, j) +0.5870G (i, j) +0.1140B (i, j), wherein f (i, j) represents a gray-scale value, R (i, j) represents a red component value, G (i, j) represents a green component value, and B (i, j) represents a blue component value;

(2c) performing median filtering on the image subjected to gray level transformation by using g (i, j) ═ mean { f (i-k, j-l) }, (k, l) ∈ H, wherein (i, j) represents a position coordinate of a certain pixel of the original image, g (i, j) represents a pixel value after gray level transformation on the position of the original image (i, j), f (i-k, j-l), (j, l) ∈ H represents each pixel value of a filtering template window H on the pixel of the original image (i, j), and mean { } represents a middle value of all pixel values in the selected { };

(2d) using | G | ═ G for median filtered pictures_x|+|G_y|，

Performing edge detection, wherein A represents the original image subjected to median filtering, G represents the image subjected to edge detection, and G_xRepresenting the gray value of the image subjected to horizontal edge detection, G_yRepresenting the gray value of the image detected by the vertical edge;

(2e) using P-P for picture subjected to edge detection₁₁&P₁₂&P₁₃&P₂₁&P₂₂&P₂₃&P₃₁&P₃₂&P₃₃Performing corrosion calculation, wherein P is the central pixel point of the corrosion calculation, P₁₁-P₃₃Corrosion calculation of 9 pixels in a central pixel 3 x 3 pixel matrix;

(3) connecting the FPGA chip with a DDR3 SDRAM, and using the DDR3 SDRAM chip as a frame buffer module for storing image data to be detected and standard image data;

(4) storing the standard images of the conveying belt recorded in advance under a good illumination condition into a frame cache module before starting detection;

(5) simultaneously extracting a standard image and an image to be detected from a frame buffer module, taking the standard image as a background frame, taking the image to be detected as a current image frame, and using the two images

Performing a background difference algorithm, wherein X_ijBeing pixel points of the current image frame, Y_ijThe theta is a return value obtained by a background difference algorithm;

(6) judging whether the return value theta obtained after subtraction is larger than a tearing threshold value, if so, judging that tearing occurs, and if not, waiting for the next frame of image to be detected and the standard image to carry out differential processing;

(7) if the volume is judged to be 'tearing', comparing theta with delta, marking by using a marking machine if delta is larger than theta, and if delta is smaller than or equal to theta, emergency stopping and giving an alarm, wherein delta is 'tearing' the volume threshold;

(8) and transmitting the picture judged to be torn to a display through the HDMI.

Images