The Research On Mini-size Precision Measurement Based On Image Processing

The development of measurement technology indicated the level of science and technology to a certain degree in one country. In the present, with the rapid development of production,science and technology,the higher demand was put forward to the measurement precision , the measurement efficiency and the measurement automation of tiny dimension. So studying on the improvement of the precision of tiny dimension's measurement has practical significance. This paper has studied the method on how to improve the measurement precision of tiny dimension from software algorithm based on image measurement technology.Digital image's conventional treatment is the basic theory of image measurement technology. The paper analyses noise source and property in digital image, and presents common filtering methods to suppress noise, what is more, it emphasis on the methods of edge inspection and image threshold segmentation .In the aspect of edge detection, first-order differential edge detection operator, second-order differential edge detection operator, and canny edge detection operator are studied and discussed. In the aspect of image segmentation, histogram method, optimal threshold segmentation method, P parameter method, and maximum variance threshold method are mostly discussed. Through the experiments, it compares the advantages and disadvantages of the edge inspection methods mentioned above in the aspects of inspection time and result.The technique of sub-pixel edge orientation is an efficient method which uses software arithmetic to advance measure precision, so it plays an important roll in image processing. According to characteristic of mechanical part, an improved orientation method used to line edge-normal least-squares linear regression method. The method changes requirement of the error squares sum in vertical direction is the minimum into taking minimal sum of squares of error in normal direction as the goal. On the premise of ensuring speed, the method reduces sub-pixel fitting allocation error, and increases measurement precision. Here it presents dimensionality reduction gray-scale moment method and improved least-squares method. The experiments indicate that the method presented in the paper calculates faster and gets higher precision.