Segmentation Of Cells Image Based On Mathematical Morphology

In recent years, medical images play an increasing important role in improving the diagrosis precision and turn into the doctor's assistant. Now ,in our country image diagnose is depanding on observation and clinical experience of the doctor. But a mass of fuzzy edge of the medical image not only waste so much of time and energy, but also difficuty to receive correct diagrosis ,further more ,perhaps result in leaking and missing digarosis. So medical image's CAD (computer– aideddiagnosis , CAD)process important studying and applied value. In CAD systems,the correctness of cell automatic recognition is concerned with the diagnostic correctness. The grayscale of some cells is similar to background noise and edge illegibility in the medical image,which become the puzzle of the madical image segmentation.In madical image segmentation ,the main methods include Histogram Algorithm, edge detection Algorithm ,Regional Growth Algorithm, clustering Algorithm ect. Most of them are aimming at concrete problems with no universal segmentation theory that fits all images.In addition ,there is no established criterions are used to select suitable segmentation algorithm , which bring to many actual problems on the application of image segmentation technique.A new method is proposed for automatic segmantation of cancerous medical image. based on morphology edge and Ostu Algorithm. That is by the witch the edge of images are sharpened through open edge detecting algorithm of mathemedical morphology ;the image are smoothened by neighboring average and median filtering ; the image are segmented automatically using the improved discriminance analytical method. The method proposed has been applid to 20 cancerous medical images ,experimental results demonstrate its validity.