An optimization algorithm of image segmentation suitable for fingerprint identification ASIC

Abstract

In this paper, an optimization algorithm of image segmentation for fingerprint identification ASIC(Application Specific Integrated Circuit) was presented. Firstly, the segmentation based on point-pixel was optimized aiming at image gradient, then the segmentation based on block-pixel was optimized for the mean and variance of image gray-scale, lastly, the second segmentation was optimized based on frequency calculation. The computation was reduced progressively by levels. It is suitable for parallel processing by less increasing the calculation. It was proved that the identification accuracy was improved, and the image of poor quality, such as black picture, was effectively segmented by experiment. This algorithm is benefit for fingerprint identification ASIC.