Chinese Chess Character Recognition with Radial Harmonic Fourier Moments

Radial harmonic Fourier moments (RHFMs) are invariant to translation, rotation, scaling and intensity, which own excellent image description ability, noise-resistant power, and less computational complexity. In this paper, RHFMs have been applied to the rotated Chinese Chess character recognition, which is the key step in chess recognition for vision system of Chinese Chess playing robot. In order to evaluate the efficiency of this method, experiments on both toy images and real chess images were carried out respectively. The experimental results indicate that the proposed method achieves an average recognition rate of 99.49% in artificial datasets and 99.57% in real-world datasets. The results demonstrate that the RHFMs have excellent performance in rotated Chinese Chess character recognition.