Image Mosaicing Techniques Study In The Entire Imaging Measuring Method

With the development of computer techniques, image measuring techniques is widely used in metrology. But because of the conflict of precision and the field of view, when doing test on parts especially microparts, several mechanical actions must be done to complete one test. When doing test on complicated features such as contour, the test process is very troublesome and difficult to achieve high precision. To fix the problems above, this thesis put forward an idea named the entire imaging measuring method and studied the application of image mosaicing in this idea. It is a good way to succeed normalization, automation and intelligence in test process. This thesis's main work are:1. Considering the shortages of presented image test method, put forward an idea named the entire imaging measuring method and study means to achieve this idea.2. Widely study development situation, priciples, characteristic and key techniques in image mosaics. Have done some experiments to validate major methods of it.3. Analyse the process of image mosaicing. To enhance image quality, study greylevel correction and achieve a good result. Analyse and compare statistic based image registration such as phase correlation and template correlation and feature based mothod and study subpixel registration. Also, do some work on a corner match based micropart mosaicing method.4. Study and use subpixel techniques to achieve high-precision feature location. Analyse the principle of image formation to find feature locating standand and find out the best way for edge location. Make use of space moments, Zernike moments and curve interpolation techniques to locate subpixel edge. Do some works on 1-dimension cubic spline interpolation combined with differential coefficient and direct least square fitting of ellipses to achieve subpixel precision.5. Bring forward a benchmark template based mosaicing method, make use of benchmark template to add outside information in micro-vision effectively, and achieve high precision feature location, make test process achieve normalization, automation and suitable for industry application.