Study On The Figerprint Compression Algorithm Based On Characteristic Of Texture

Fingerprint image compression is an important component of automated fingerprint identification system(AFIS) due to the increasing number of the fingerprint records in their databases. In this paper, we propose three fingerprint image compression algorithms based on characteristics of fingerprint image and high performance algorithms such as SPIHT, MRWD, SLCCA, JPEG2000.(1)Fingerprint compression based on SPIHT and ROI code of minutiaes in the fingerprint image. (2)Fingerprint compression based on directional image and morphological dilation.(3)Fingerprint compression Based on directional image and Significance –Linked connected component. The first Compression algorithm which is based on Lossless or nearly Lossless Region Of Interest(ROI), means to compress interesting regions–the minutiae in an image without loss or with less loss,and to compress uninteresting regions with loss. Thus, desired high fidelity image information is acquired while remaining a high compress rate. The second algorithm which is based on directional image and morphological dilation, using different structuring element for morphological dilation according to different directional blocks to extract and encode the clustered significant coefficient in the different subbands from low frequency to high frequency . The last algorithm which is based on directional image and SLCCA algorithm, not only exploits within-subband clustering of significant coefficients but also cross-subband dependency in the significant fields. It seeks to enhance the second agorithm by building the so-called significant-link between a parent cluster and a child cluster across-subband when two connected component if the significant parent belongs to one component, and at least one of its children is significant and lies in another component. If the positional information of the significant parent in the first component is available, the positional information for the second component can be inferred through marking the parent as having a significant–link. And a significant saving on encoding cluster positions is thus achieved. Apparently, marking the significant–link costs much less than directly encoding the position. Compared to WSQ algorithm PSNR improvements up to 1.0 dB are achieved in the new algorithm3 for extensive computer experiments on fingerprint, and both the encoding and decoding procedures are fast.