Research Of Image Contour Extraction Method Based On Simulation Of Biological Cells

The image edge is one of the basic characteristics of the image in the complexbackground image, which contains of most information of the image. In the image edgedetection, contour image is a very important feature, so the outline image featureextraction is one method of the important feature extraction, too. Contour extraction isdiffident of the edge detection, which can remove the background texture information atthe same time of extracting the edge of the image.Human visual information processing mechanism of the human is the main way ofour brain to get the outside information, which almost has a perfect informationprocessing ability, and at the same time, it has a better way than the informationprocessing of computer vision. So people are constantly looking for a mathematicalmodel which is close to the human visual system, using it to simulate the function ofvisual information processing mechanisms. Fortunately, the human visual system caneffectively extract the contour feature. In the primary visual cortex, the outer peripheryof the classical receptive field has an area of classical receptive field inhibition inresponse to external stimuli exist, which is known as non-classical receptive field. Thisinhibition of the target contour extraction from complex background texture plays animportant role.This thesis describes the mechanism of visual contour extraction from the twomethods. The first method is based on the non-classical receptive field smoothingbilateral suppression characteristics of dynamic contour extraction. First, the bilateralfiltering method for image smoothing, at the same time, it retain the edges. Secondly,with the help of orientation and the relative position of the dynamic characteristics,correct the non-classical receptive field concentric model weights. With the help ofsimulation, the model extracts the target contour, and at the same time, it removestextured background. The second method is to improve the non-classical receptive field toward smoothing contour extraction based on select properties. The method is based ona method for smoothing areas were improved, which makes the simulation has beenwell validated and receptive field properties so that applications in computer visionpossible.