Technology Research, Based On The Edge Of The Blood Cell Image Segmentation

With the development of the disciplines and industries of science, technology, agriculture and medicine in recent decades, there are more and more signals and images needing for processing in production and our lives, and the amount of work is increasing rapidly .In this instance, the automatic technology of signal processing is needed badly and the development of computer science and information technology makes it available .Nowadays ,digital image processing technology has been used in many fields of industry and agriculture and many practical image processing theory come into being in consequence, medical image processing technology included.Taking advantage of theories and methods of image segmentation and wavelet transformation and according to the structure characteristic of blood cell image in microscope, I initiate series of theories and technology that cater to this kind of image and produce a suit of practical automatic detecting software systems of blood cell .This article discusses the corresponding theories and algorithm of image technology and gets some experimental results.Because of the differences in coloring agent materials and shooting circumstances, the image will be different from one another .However, there is one feature in common that a hole seats in the inner of the cell, which is caused by blood cell's structure characteristics and chemical composition .It is difficult to get a satisfying result if we only use region continuity-based image segmentation method. Therefore, the paper puts forward an edge-based image segmentation method and gets preferable results. Besides of the segmentation methods, I focus on wavelet transform and work out a suitable wavelet modulus, which is used in hormomorphic filtering and edge detection.The thesis discusses the problem of noise removing and background rectifying firstly. Although many image de-noising methods do well in reducing the influence of noise, some edge information is weakened as well, which is not beneficial to the work of edge-based image segmentation. Therefore, the paper designs de-noising methods with selectivity and gets good results and has a discussion on two kinds of hornomorphic filter methods on backgrounds correcting.The discrimination of object from background is the chief content of image segmentation; it is also the weight-bearing point in this article. Because of the inadequacy of traditional region-based image segmentation method in the field of blood cell image segmentation and according to the distinctive edge characteristics of different blood cell image, the paper build up a suit of theory and algorithm of edge detection and processing based on which the paper reaches the aim of preliminary image segmentation .The experimental results prove that the influence of holes that seats in the center of the cell to segmentation and measurement is well avoided .The segmentation of adhesion cells is another important component .Adapting mathematic morphology, the thesis first extracts a series of seed regions from the pre-segmented image by the way of erosion and makes the adhesion cells separated by conditional inflation. Then I take a measurement of the segmentation results and get some parameters of the cells, including circumference, square, major and minor axis.