Research On The Automatic Identification Of Particles In Microscopic Cell Image Based On Machine Vision

Microscopic cell detection is a routine measurement in medicine, it is a key step that how to identify the particles in microscopic cell image quickly and effectively, so it has a significant meaning in medicine to research the fast automatic identification of particles in microscopic cell image. In recent years, with the rapid development of machine vision research and the improvement of the biomedical image processing technology, microscopic cell image processing technology in medicine has been gradually changed from the traditional manual identification to the automatic identification by computer. Adopting image processing technology in machine vision as its core to the automatic identification of particles in microscopic cell image has become the emphasis research in the literature. This paper mainly studies the process of particles in microscopic cell image from the acquisition to the identification based on machine vision. The main contents are as follows:1. The hardware and software structure of the automatic identification system based on machine vision are emphasis discussed. The hardware structure includes the design of the lighting system, the selection of the CCD camera and image acquisition card. The software structure mainly contains the acquisition procedure and the process flow of the microscopic cell image.2. According to the gray-scale characteristics of the microscopic cell image, the method of image preprocessing is studied, a better filtering method is obtained by contrast experiments. By studying and comparing the traditional segmentation methods, the optimization segmentation method based on the 2-D maximum entropy threshold combining morphological operations is proposed. The method can segment the particles area from the microscopic cell image effectively. The segmentation method is more accurate and fast.3. The features of several common types of particles in microscopic cell image are in detail introduced, and the feature extraction method is also studied. Then a set of feature vector is appropriate determined and regarded as the input of particles in microscopic cell image according the contrast. It can identify the cell features usefully and represent the characteristics of all kinds of typical particles in microscopic cell image.4. The theory of SVM(Support Vector Machine) and BP(Back Propagation) neural network are introduced, and the classification algorithm based on support vector machine is emphasizes studied. According to the features of particles in microscopic cell image that extracted before, an automatic identification classifier of particles in microscopic cell image based on support vector machine is designed. The classifier is trained and tested by the microscopic cell samples and the experiment results are ideal. At the same time, by the comparison results of classification of BP neural network classifier which shows the obvious advantage of SVM.Therefore, the automatic identification techniques based on machine vision studied in this paper are feasible, effective which has a certain practical value in medicine.