Research On Key Technologies Of Cell Microscope Image Analysis

Cell microscope image analysis technology is a key issue of biomedical engineering field domestic and abroad in recent years. This research is mainly concerned with those methods which adopt computer vision technology to replace the human observation process, in order to carry out automatic analysis of microscopic image accurately and efficiently.This paper focuses on four key points of cell microscope image analysis technologies, they are object location, segmentation, registration, and recognition, which were deeply discussed and specific implemented. In this research, the main tool is the theory of digital image processing and pattern recognition, which combine with various kinds of real word problems to propose specific algorithm of actual applications. The main innovations of this research can be summarized as follow:1. Proposed an algorithm of adaptive threshold segmentation based on color space conversion and histogram gradient for leukocyte location, which achieved the automatic location of normal leukocyte in the microscope visual field.2. Proposed an algorithm of color space region growing segmentation based on seed points of mathematical morphology, which achieved a more accurate structural segmentation of leukocyte.3. Proposed an algorithm of rigid registration based on centroid and phase correlation for plant embryo section images matching, which achieved the translation and rotation parameter of corresponding section image.4. Proposed an algorithm of mathematical morphology segmentation based on scale space and watershed for granule extraction, and designed a classier using BP neural network and some features, which achieved the granule extraction from the net-like cell wall structure in the microscope visual field. According to the special requirements of different applications of image analysis, a large number of experiments were carried out. The results verified that these methods were effective to the corresponding applications.