Research On Precision Measurement Technology Of Image Processing

Image measurement is a new technique in measuring field. It is widely used in delicate parts, dimensional measurement of geometrical quantity and remote-sensing measurement of avigation. The principle of image measurement is detecting the edge of an object by using digital image processing theory to obtain its geometric parameters. Currently, with the rapid development of production, science and technology, higher demand is put forward to the solving precision and the efficiency of tiny dimension.Based on software processing image measurement techniques, this paper introduces hardware and software design of image adjustment process, presents the design and implementation of a high accurate measurement system in terms of image handling. It focuses on the key technology to improve the efficiency and accuracy of the process.Considering practical requirement, the author first analyze our algorithm from the perspective of SUSAN method, and compare it with traditional ones. The comparison shows that this algorithm improves the precision and speed of edge detection;We then develop an image processing software based on Visual C++ 6.0.After that, we propose a new edge detection algorithm, which combines method of Segmentation method to select the smallest error and configuration extraction to achieve the detection of real edge of the target image ,and used MATLAB to compare with other common edge detection method. In the end, based on the theory of least-squares method in sub-pixel edge orientation, this paper propose BP(back propagation) network, which use to achieve edge fitting,The simulation results show that, after 234 times training, the fitting error is less than 1.0×10?6.