Research Of Coal Level Image Preprocessing Method Based On Auxiliary Light Source

The mine monitor system is an important facility that ensures the mine to operate safely and efficiently. The detection of coal level in underground coal pocket is important portion of the mine monitor system. Though there are various detection manners, they are many shortages and short lifespan. The special collecting image manner is given. In order to exam the coal level real time, the detection of coal level in underground coal pocket is possible in using image recognition technology. Image acquiring method and the image preprocessing method is the important guarantee of the coal level image effective recognition. This article has mainly conducted following research: (1) According to the establishment characteristics and the working conditions of the coal bin, this article has summarized the coal bin image characteristics, and studied how to use auxiliary light source to acquire the coal level image. Using the image characteristic which produces on the coal surface to exam the coal level, it not only has effectively overcome the influence of each kind of disadvantages, but also realizes simply. It suits to exam coal level. (2) This article in view of luminous intensity weakening, the background disturbance, as well as disturbances by the mechanical vibrations and the coal dust, uses the corresponding enhancement and the restoration processing method to make it sharp. To the images weakened by the background disturbance and the luminous intensity, we first use gradation transformation to enhance the images. Then we use the wavelet transformation and morphological reconstruction to process the images. To the images weakened by mechanical vibrations and the coal dust, we use the constrained least squares method to restore the images. First we estimate the expansion function, and then use the constrained least squares method to restore the images. After processing, we can obtain the sharp images. (3) After preprocessing the images weakened, this article has used the ant colony algorithm to detect the edge of images. First, we fix the cluster center, and then use the cluster algorithm to detect the targets completely.