The Research Of Image Pretreatment Algorithm Based On Material Stack Measurement System

In order to meet the needs of large enterprise logistics information management, it needs to solve the effective measurement of material stack which storages in the outdoor yard during the management process for raw materials. Based on the analysis of several existing three-dimensional measurement method, this paper make deep research in the method of binocular vision measuring the volume, and ultimately come up with a new scheme:project the black and white grid with special information to the surface of material using a point light source, and install two cameras around material stack, and then shoot the material stack from different direction. At last send the image information to computer, the computer will do image pretreatment, camera calibration, feature point extraction and matching,3d reconstruction, then the volume of material stack can be quantitatively calculated.First, this project designs an experimental device, and then obtains two pieces of RGB digital images that contains material space information. Meanwhile, study the camera calibration and select the direct linear calibration method to get projection matrix of the two cameras, and also can continue unfold the projection matrix to obtain the inside and outside parameters of camera. This method is simple and the requirement of the calibration object is not strict, so it is suitable for large stockpile volume calculation.Second, preprocess the material image including image contrast enhancement, feature extraction and matching. This paper use grayscale of color image, histogram equalization and image denoising to achieve the contrast enhancement. Based on the research of feature extraction and matching methods, this paper selects the artificial registration method to extract the matching points of material image.Finally, based on the inside and outside parameters of the two cameras and matching points, the paper realize three-dimensional reconstruction of feature points according to mathematical model, and then obtain three-dimensional coordinates of space point. Finally, according to three-dimensional coordinates, the volume of material can be calculated.To some extent, not only does this method of volume measurement achieve the automation of measurement, and also ensure the accuracy of experimental results through laboratory testing. The method is simple, low labor intensity, low cost, suitable for general spread.