To encode an image by fractal image coding, it needs to find the best matched domain block for each range block. The best matched criteria is the difference measure between the range block and the domain block transformed by affine transform. This paper proposes a new method of fractal image coding, called weighted fractal image coding. To enhance the visual quality of the decoded image, the edge regions of the image are encoded with high weighting. The contrast scaling factor and the luminance offset factor in the affine transform are derived by minimizing the weighted difference measure. Experimental results show that the proposed method improves the visual quality of the decoded images in terms of weighted peak signal-to-noise ratio.

In this paper, a fuzzy normalization method is proposed for normalizing Chinese character images in preprocessing stage of the Chinese character recognition. It is known that the variations of handwritten Chinese characters are great. If we can reduce the variations and preserve the main features of characters in preprocessing the handwritten Chinese characters, the higher recognition rate will be achieved. The new technique of the proposed fuzzy normalization can effectively deal with the irregular length of strokes, especially at the starting or ending stroke of the Chinese characters. By intensive experiments, we compare the new fuzzy normalization method with the general scaling method, and the results show that the proposed method performed well in Chinese character normalization and achieved better recognition rate.

In the literature, m/sub j/-projection has been applied for residue error detection and correction. This brief contribution extends the m/sub j/ projection to K-term m/sub j/-projection so that single residue error correction can be accomplished in an efficient manner. Based on K-term m/sub j/ projection, the necessary and sufficient conditions for the correction of single residue digit error are derived and an efficient procedure for single error correction is given.>