Research On Key Technology Of Processing Scanned Images Based On Micromanipulation Platform

Microscope is an important instrument to recognize the micro-world for human beings.In the past, it depended on adjusting by hand to observe something under a microscope, which costs significant time. As technology develops, combined with computer vision and intelligent control technology, microscope automation is done gradually. According to the practical demands for application of micro-operation platform, resetting and image mosaic in big vision are researched in this paper.A platform with functions of auto-control, auto-scanning and storage is designed firstly,especially in the latter a method of micro-images arrangement and storage which can be used for further analysis is used here.With regard to the problem of microscope resetting, a method for coordinate determination is proposed using label. Hough transform is used to identify circles of labels. It works by connecting centers of two circles to form a vector, which is then used to determine the coordinate of PC screen. In the process of automatic calibration and scanning path recalling, by changing among CCD coordinate, screen coordinate and platform coordinate,one can return to the region interested according to relative paths by control motors and the image will display right in the middle of the screen. This function can also be applied to auto recalling of scanned images. It can guarantee the precision of repeating positioning and cost less on platform at the same time.As to the problem of image mosaic, several popular image registration methods are compared, and SIFT algorithm is chosen in the end for its higher stabilization. Besides,detailed deductions and corresponding information of images about micro-image registration is given. Before image fusion,detailed descriptions of work on coincided regions after micro-image registration are provided. This laid a solid foundation for later image fusion by weighted fusion Gaussian model. At last, as registration and fusion of micro-images to be used for coinciding of big regions and small regions, the stitched image has good seamless and smoothing characteristics.For the demand of observation in big region under high resolution on the micro-operation platform, a method for fusion of several micro-images stitching is given.Afterward contour recognition and statistics designed of blood cell images are done, which makes the key technology of micro-operation platform applicable.