Research Of Micro-parts Precision Testing Based On Whole Image

With the development of computer techniques, image measuring techniques is widely used in the inspection of industrial parts. However, in the image measuring of microparts, the imaging area of the measuring system is reduced to obtain high measurement precision. When measuring a number of points on the micropart, the frequent relative movements between the imaging system and the micropart are needed to obtain the image data. So the measurement with low efficiency can hardly meet the demand for the comprehensive inspection of the volume key parts in industrial field.To obtain efficient, precision and automatic inspection, this thesis studies the edge detecting and tracking technique, discusses and evaluates the subpixel location algorithms. The main research work in this thesis includes:1. The image filter technique, traditional edge detection arithmetic operators and the mathematic morphology erosion method which aims at the edge extraction of the binary image are studied in this thesis. Present intelligent edge tracking method that automatically tracks and saves the contour data in the image.2. Discusses the cause of the fuzzy edge and the two methods of subpixel location of the edge, spacial moment method and interpolation curve method, that derivative from the definition of edge. Base on evaluating and concluding some subpixel algorithms, a subpixel algorithm that cubic spline interpolate along the grads is presented and improved by non-isometric spline interpolate along the normal.3. Founds a algorithms assessing system based on composite image. Investigate the influence of edge distilling algorithm, point spread function, background optical intensity and noise on subpixel algorithms.4. Discuss the methods to make the subpixel algorithm more accurate in measuring the parts. Investigate the postural regulation means of the parts. Measured the profile error and the key parameters.