A novel fingerprint orientation field estimation method

作者：Liu Yuyu^*; JunGu; Liu Ming

来源：3rd International Symposium on Computer, Communication, Control and Automation (3CA 2015), 2015-08-09 to 2015-08-10.

• Fingerprint recognition
• Orientation Field
• Binary-pattern
• Global approximation

As a global feature of fingerprint pattern, the orientation field plays an important role in fingerprint recognition system. Most of orientation field estimation algorithms are based on gray scale local gradient calculation and averaging. However, these methods are computationally expensive for embedded environment and their results are sensitive to noise, especially in low quality fingerprint images. In this paper, a binary-pattern based orientation field estimation approach is proposed. The proposed method estimated local ridge gradient in binary images instead of gray scale ones, which significantly reduces the computational complexity and increases stability of local ridge gradient. In order to eliminate the noise effects, a global approximation is applied to the estimated block-level orientation field. Experimental results show that the proposed approach has a robust performance on different fingerprint images, and its low computational complexity satisfies the strict limitation of embedded system. Extensive experiments show that the overall performance of fingerprint recognition system can be improved by applying the proposed method instead of the gradient-based ones.

• 出版日期2015
• 单位大连海洋大学