Research Of Processing And Compression Based On Dark Field Image

With the development of technology,the detection of optical component is becoming much more important and automation degree is much better.Combining machine vision,Automated Optical Inspection is becoming the tendency of inspection industry.And image processing occupy core status.This paper aims at image processing based dark field image acquisition and the main contents of this paper are as follows:Image acquired from linear scanning system may combines vibration,and this paper proposed an algorithm to remove the vibration.By analyzing the cause and harm of vibration,we proposed an algorithm based on template matching.The template is made by the component itself.In order to match the template,we need to fit the center of the component,and overlap the fitting center with the template center.Then by translating and interpolating the pixels to make the algorithm happen.And we repeat the experiment many time to make sure that the algorithm is correct.Combining with spatial index structure,wo used data structure to locate and store the defects on the surface of optical component.The structure is named quadtree structure.The quadtree structure is used to segment dark field image into several levels and stores each level information into different tree nodes.We improve the traditional quadtree structure by analyzing the theory and get the improving quadtree structure.Then we introduced how the improving quadtree structure is created and how to use it to store the information of defects.With the quadtree structure we proposed a mixture compression coding,that is global Run Length Coding and local Huffman coding.Then we discussed the reconstruction of image,and evaluate the image by Peak Signal to Noise Ratio algorithm.