Contour extraction based on GVF snake model with phase preserving denoising correction

This paper proposes a contour extraction method based on Gradient Vector Flow snake model corrected by phase preserving denoising. To reduce the noise influence and preserve the image features, the Log Gabor wavelet is used to correct the gradient vector flow snake aiming to solve the confusion problem in snake model. The proposed scheme is applied to ultrasound images to determine the object boundaries and the satisfied results is obtained even in low SNR level. Comparison results and performance analysis are given which shows that the method can remove or weaken false edge and make the snake contour detection more accurate.