Transfer Belt Crack Detection Based On Image Processing

In many large-scale coal plant, Rubber belt conveyor is widely used as coal handling equipment. The conveyor belt is the most important coal carrier, once the belt is torned that will not only bring significant safety risks, but also bring huge economic losses to the production units. Therefore, early prevention of tear tape and detection of small cracks appearing on the conveyor belt have important theoretical and practical engineering significance.This paper analyzes the practical applicative environment of the coal conveyor. The calibrating model based on coplanar object is designed to CCD camera that will capture the image of the crack on the conveyor, to achieve calibration of the internal and external parameters of the CCD camera. That is ready for the collection of the conveyor cracked image.There are low image contrast, noise and other issues in the conveyor cracked image. The first, please do image enhancement to the collected cracked image. On this basis, we bring forward two-dimensional maximum entropy fast thresholding algorithm in real-time more improved. Through the neighborhood average gray value of the one-dimensional gray level histogram to obtain the maximum and minimum gray value to compress the looping scan range from the original entire gray area to between the minimum and maximum gray value. This will shorten the running time and better meet the real-time that the system requirements.Against the cracked image may exist fault conditions after the image segmentation, and the cracked image itself may exist the case of overlapping, we propose a method based on mathematical morphology. By constructing the corresponding structural elements in different directions to separate the overlapping cracked image. By constructing the structural elements in both directions of left oblique and right oblique 45 degree to bridge the cracks on the fracture, and crack bridging recovery after, Proved to bridged and recovery the cracked image and the original binary image have the length and width of infinite approximation. That is the preparation for the accurate calculation for the characteristic parameters of the cracked image.Regardless of the direction of cracks, the method calculated parameters of the area region is relatively simple. However, due to the different definition of length and width of different the direction of the crack, we put forward that from the mathematical point through the establishment of coordinate system and rotating the cracked image to calculate the length and width of the crack parameters. The any direction's crack is easily find the length and width. Experimental results show that the algorithm has good adaptability and high accuracy.