Rotation-Invariant Pattern Recognition of the Multiple Circular Harmonic Filter Using Proper Center

적정의 중심점을 이용한 다중 원형 고조 필터의 회전 불변적 형태 인식

For the rotation-invariant pattern recognition, we propose multiple circular harmonic filter which is expanded about the proper center. The proper centerm when input image is given, is the circular harmonic expansion center of the filter which yields a maximum center correlation peak in the output plane. In this paper, we founded the circular harmonic components that the proper center superposes on the same position and then designed multiple circular harmonic filter using these components. Also the proposed filter is compared with conventional multiple circular harmonic filter and shows that it can maximize the center correlation peak for the rotated input image by the computer simulation.