Knowledge based Palm Lines Extraction with Adaptive Thresholding

M. Wong and D. Zhang (PRC)


Adaptive thresholding, palmprint, principal lines, wrinkles,ridges, knowledge based, line feature extraction.


There are three types of line features from a palm: principal lines, wrinkles and ridges, which have different directions, width and intensity. Traditional edge detection techniques cannot extract the palm lines effectively. In this paper, we propose a new method on the line extraction from a palmprint image by incorporating knowledge. We first use adaptive thresholding method on the binarization of palmprint images, and then we define the bifurcation points and end points in a palmprint image to create a palmprint structural map. Next we establish some searching strategies on exploiting the structural information of line segments from the map. We obtain: 1) regular bifurcation, 2) short bifurcation, 3) loop, 4) bridge, and 5) regular lines. After that, decision on removing or joining line segments can be conducted to form the final palm line features. Our preliminary results indicate that by applying prior knowledge with adaptive thresholding on the palmprint image, we can get the major lines effectively.

Important Links:

Go Back