Image Segmentation Using Proportion of Foreground to Background Algorithm

The medical image is widely used in medicine diagnosis for the science-sound qualitative and quantitative analysis. To achieve a clear image, the technology of the image segmentation is the key point. Therefore, this paper proposes a new image segmentation approach using the proportion of foreground to background method. Based on the traditional grayscale image segmentation and the edge detecting method, the edge M1 of the original image is firstly extracted by Sobel operator templates in 8 directions. Secondly, the proportion of foreground to background in image M1 is calculated to find the threshold T in the iterative loop. Finally, the segmentation M2 is thus obtained according to the threshold T. Experiments with the human brain image indicate that the number of iteration steps for the threshold T is reduced. Also, the phenomenon of over-segmentation and the discontinuity from the traditional edge detecting algorithm can be avoided. The experimental results are provided to verify the proposed approach.