Research on image segmentation method based on improved Snake model

Image segmentation is one of the key research fields in computer vision, and the research of image segmentation methods based on active contour model has been continuously advanced in recent years. Aiming at the defect problem such as traditional Snake model algorithm is more sensitive to the noise of the original target image, it is proposed that an improved segmentation algorithm based on bilateral filter to replace the original Gaussian filter of the traditional Snake model, to reduce the noise of the original target image, by weighing the spatial domain weights and domain weights of the pixel points, so as to achieve the purpose of edge denising, so that the original target image edge contour can be further optimized and extracted; By using the snake model before and after improvement, we performed a qualitative and comparative analysis for the extraction effects on edge contour of the same original target image object, and it was verified that the improved snake model proposed here is more accurate and effective. The accuracy and effectiveness of the improved model here are objectively and quantitatively verified, according to the number of sampling points extracted, peak of noise-signal ratio(SNR) of the result map extracted and image quality of original target image object edge profile.