基于方向图及小波系数重要连接的指纹图像压缩算法

提出了一种在SLCCA算法基础上结合指纹方向图的指纹压缩算法。首先将指纹图像按纹理信息进行分块，然后根据各指纹图像方向块的方向，确定不同的结构元素对小波图像各子带中的重要系数进行数学形态学的膨胀操作来提取各子带中的重要系数簇。再根据不同予带间重要系数簇中父系数与子系数间的重要系数分布建立重要连接。算法不仅充分考虑了小波分解图像不同子带内的重要系数分布的聚类特性，同时也充分考虑了各子带间的重要系数分布的自相似性。实验结果证实该算法不仅比WSQ、EZW、SPHIT、MRWD、SLCCA算法具有更好的性能。而并且可与最新的有损压缩算法JPEG2000相媲美。同时，编解码的速度较快。

Fingerprint Compression Algorithm Based on Directional Image and SLCCA

The paper proposes a fingerprint compression algorithm which is based on directional image and SLCCA algorithm. At first it divides fingerprint image into several blocks according to their directions, and then uses different structuring element for morphological dilation according to different directional blocks to extract the clustered significant coefficient in the different subbands from low frequency to high frequency. Then, it uses the so-called significant-link between a parent cluster and a child cluster across-subband. The proposed algorithm not only considers within-subband clustering of significant coefficients but also considers cross-subband dependency in the significant fields. Extensive computer experiments on fingerprint images show convincingly that proposed algorithm outperforms WSQ , EZW, MRWD, and SPIHT in PSNR, and the encoding and decoding procedures are fast. Besides, it provides the comparable performance to that of the JPEG2000 compression algorithm.