Detecting image symmetry based on phase information

Traditional methods for detecting symmetry in image suffered from the contrast of image and noise, and they all needed some preprocessing; for images including many objects, they must segment them to detect the symmetry of each object. This paper presents a new method of image symmetry detection. Instead of gradient information, it used phase information of original image. Phase information is stable and significant, so the method is robust; it is invariant to contrast, rotation and illumination conditions. It can detect mirror symmetry, rotating symmetry and curve symmetry at one time and does not need any preprocessing, such as segmentation. Results of experiment has shown that compared with pivotal element algorithm based on intensity information, this method is more accurate and robust. This method can be used in pattern recognition and image processing; it is also very useful in location and recognition of object and rebuild of 3D objects.