Abstract: To overcome the limitations of traditional snake image segmentation model， an improved dynamic directional gradient vector flow (DDGVF) based on wavelet transform was proposed. First， the image to be segmented was decomposed into 3 layers by using the multiscale analysis of wavelet transform， then the DDGVF segmentation was performed under each layer of the decomposition of the image， and finally， a more accuracy target contour could be acquired. Compared with the other image segmentation methods， the proposed algorithm can better segment the depression area of the target image， get a wider capture range and spend less time. The effectiveness of the improved algorithm has been proved through the simulation experiment of the synthetic image and the real medical image.

Key words: medical image segmentation, parametric active contour model, wavelet transform, dynamic directional gradient vector flow (DDGVF)