Mahjong image recognition scheme using Fourier transform technique

In this paper, we analyze the Mahjong pattern and divide the pattern into several sub-blocks for extracting features by DFT. In normalization stage, scaling method is used to standardize patterns so to eliminate image zooming in or out problems. We use two strategies to improve the Mahjong image identification: (1) we develop a special normalization skill to standardize the image. (2) We develop five features for pattern identification. The features can exactly distinguish which is the test pattern chess. Compared with other methods, our scheme has simple and effective advantages. In order to demonstrate the effectiveness of the proposed scheme, simulations under all kinds of various conditions were conducted. The experimental results show that our proposed scheme can exactly identify Mahjong pattern images at 100% accuracy all the time.