Study on Application of Computer Vision Technology to Pneumatic Proportional Position System

Abstract

The paper combines computer vision technology and pneumatics, and designs pneumatic proportional position system based on computer vision technology. In order to realize automatic location of pneumatic system actuator, the paper studies preprocessing technique of image including gray scale conversion of image, image enhancement, smooth, improving image quality and highlighting objective features. And the paper researches image segmentation approach, and applies mathematical morphology to process binary images, which makes image quality better. Introduction In recent years, with the development of computer technology, especially the development of multi media, digital image preprocess and analysis theory, and rapid development and application of large-scale integrated circuit, computer vision technology is widely applied and studied. This paper applies computer vision technology to study the location of objects, which provides the position information of the objects for pneumatic position system, for achieving automatic location of pneumatics, which not only makes pneumatic system have object recognition ability, but also constructs the system of object automatic recognition and location. The paper realizes the application of computer vision technology to pneumatic system, constructs a computer vision pneumatic positioning system, and realizes the function of object recognition and location, which provides reference for real production application. Above al, studying computer vision, especially studying the application of it to industry, has great importance to improve production technology and develop industry. Image gray processing In order to accelerate the processing speed of images, the acquired color images need to be converted into gray images, which is called gray processing. It makes R,G and B of RGB model equal. Gray processing is the process that color images with light and color are converted into gray images. The graying result is the basis of processing the subsequent images. In general, each pixel of color image is represented by three bytes. Each byte corresponds to the light (red, green and blue) of R, G and B component. The pixel of the black-and-white image after conversion uses one byte to represent the gray value which is between 0 and 255. The greater the value is, the whiter the point is. The transformational relation is: ( , ) 0.299 ( , ) 0.587 ( , ) 0.114 ( , ) Gray i j R i j G i j B i j = + + (1 ) ( , ) Gray i j is the gray value of black-and-white image after transformation on point ( , ) i j In formula (1), green has the largest proportion, so G value can be used to be as the gray after transformation. Gray processing has the other methods such as taking the maximum, minimum and average value of three components. The purpose is to make the values of R, G and B component equal. For gray-scale images, each pixel value can be represented by one byte or eight bits, which can show 256 ( 8 2 ) levels. It means that gray-scale images only can show 256 colors, and the gray-scale images only have gray level and have no color. International Conference on Automation, Mechanical Control and Computational Engineering (AMCCE 2015) © 2015. The authors Published by Atlantis Press 331 Image Segmentation and Location After the image receives gray-scale conversion, histogram equalization and smooth, the image is segmented. The objects of the image are separated from the background, which forms a binary image. Based on the image, morphological processing is made to make the centroid coordinates calculation accurate, which realizes pixel coordinates location of object image centroid. Segmentation based on regions. Selecting gray threshold for binarization processing on object image is one of the most typical, simple and effective image segmentation methods. Before taking out geometric features of gray scale images, the images must receive threshold processing, which means the process of taking out the part above or below the gray value. An acquired image includes object, background and noise. The most common method for how to take out the required object image from many-valued images is to suppose a threshold value t, and t divides the data of the image into two parts, the pixel which is greater than t and the pixel which is less than t. If the input image is gray scale image ( , ) f i j ,and the output image is ( , ) g i j , 1 ( , ) ( , ) 0 ( , ) f i j t g i j f i j t ≥  =  <  ,