Weighted Multi-scale Composite Window Fingerprint Direction Field Establishment Method and Hierarchical Smoothing Algorithm

Abstract

The invention discloses a weighted multi-scale composite window fingerprint direction field establishment method and a hierarchical smoothing algorithm. The present invention establishes a series of fingerprint direction field OF information under composite windows of different scales; then determines the weight of each scale in the whole series through the square gradient consistency index; finally, integrates the OF information under the whole series through the weight to obtain a fingerprint Orientation field OF; effectively reconstructs the fingerprint orientation field OF of low-quality fingerprint images, and better fits the direction of the singular point area, can accurately estimate the direction of the noise area and avoids the offset of the singular point position. In addition, the present invention uses a hierarchical smoothing algorithm to process the fingerprint direction field to repair the wrong ridge line structure while ensuring that the real singular point position does not shift.

Description

technical field [0001] The invention belongs to the technical field of numerical image processing, and particularly relates to a method for establishing a weighted multi-scale compound window fingerprint direction field. Background technique [0002] Orientation field (OF) shows the overall structure information of fingerprint ridges, and describes the basic shape, structural features and ridge information of fingerprints from a macroscopic level. Accurate and reliable OF has an important impact on many subsequent fingerprint image processing, such as fingerprint image enhancement, singular point detection and fingerprint image classification. [0003] The gradient method to calculate the fingerprint OF has the advantages of low computational complexity and intuition, so it is one of the most widely used fingerprint orientation field algorithms, but it is also easily disturbed by noise. In order to improve the OF calculation accuracy of the existing gradient method, the fol...

AI Technical Summary

Problems solved by technology

Subsequent researchers have improved this method to obtain a more accurate OF, but these methods are difficult to balance the accuracy and robustness issues.

Because the OF calculated by small blocks is more accurate but sensitive to noise, the OF calculated by large blocks is more resistant to noise but less accurate

Therefore, the researchers proposed the gradient method of the composite window. The experimental results show that the composite window can better balance the accuracy and robustness issues compared with the single window, but the composite window introduces a complex problem: how to choose the inner The size of windows and exterior windows? Existing methods rely on a large number of experiments to select more appropriate inner and outer window sizes based on experience

Unable to adaptively choose the size of the composite window

[0005] (2) How to accurately reconstruct the ridge information near the singular point of the low-quality fingerprint image without causing the singular point position shift? The existing gradient method has poor noise resistance for low-quality fingerprint images. In order to reconstruct the OF information of low-quality fingerprint images, researchers have proposed many improved gradient methods, such as hierarchical gradient method, weighted average method, and composite window gradient method. etc., but because of the higher curvature of the ridge near the singular point, these methods will get the wrong ridge structure near the singular point or cause the position of the singular point to shift when the low-quality fingerprint image OF is reconstructed.

When the offset singular point is used as a feature point for subsequent fingerprint matching or identification, it will cause large errors or even errors in the matching and identification results.

Images