压缩链码的研究

对图像处理及模式识别中的链码技术进行研究.首先针对具有二位码长的顶点链码只有3个码值的情况,在不增加码长的情况下对其增加了一个码值,大大提高了其表达能力.其次,同样根据顶点链码的上述冗余情况对其进行了不等长编码压缩.然后,在上述两种改进链码的基础上,根据大量的统计数据提出了一种基于Huffman编码的新的压缩链码.文中还提出了一种对链码的评价方法.最后,用该评价方法和对实际图像的实验测试对所提出的链码和4种已有链码进行了理论上和实用方面的比较.比较的结果显示该文提出的压缩链码效率最高.

Research on Compressed Chain Code

This paper introduces the research works on chain code used in Image Processing and Pattern Recognition. Firstly, considering that the vertex chain code uses 2 bits to represent only 3 kinds of elements, a new kind of elements is introduced without increasing the average bits per code. The efficiency is raised greatly. On the same observation, an unequal-length coding of the vertex chain code is presented. Based on this, a new compressed chain code coded by Huffman method is proposed by collecting mass of statistic data. Furthermore, a method for quantitatively evaluating chain codes is devised. Finally the proposed chain codes are compared, theoretically with the evaluating method and practically, with the existing four ones. The result of the comparison shows that the new compressed chain code is most efficient.