Ultrasound CT Image Segmentation Based on Region Growing
Wang Hao-quan, Xing Hao · 2010
Based on characteristics of ultrasound CT images and principles of segmentation, the largest gray value of pixels in flaw areas is taken as seed points in order to meet the pixel proximity for the criterion of region growing. Through changing segmentation direction, a new segmentation method can be obtained. The experiment verified that the method has achieved above 95% to integrity and accuracy of target areas which are small. The method can better solve the contradiction between anti-noise and segmentation precision.