Segmentation of Natural Image Based on Symmetry Features

Symmetry is one of the important features of many nature or artificial structures. In a natural image with a major subject, the largest symmetric region is usually the subject. This paper presents a method to segment the largest symmetric region in a nature image. Firstly, the image is segmented into regions with internal similarity. Secondly, according to their features, these regions are classified into similar groups or unique regions. Each group or unique region is examined to find its symmetry axis. Finally, these regions are analyzed and summarized to determine the largest symmetric region in the image. We use a butterfly image database to test our method. The experimental results show that the proposed method is effective.