Extracting region of interest (ROI) is an important process for finger vein recognition system. It is determined the part of the image that we need for extracting features. This paper presents an ROI extraction method that overcomes the problems of finger rotation and displacement based on machine learning. First is to locate the finger midline, which will be used in correcting the oblique images. Then, use a sliding window to determine the proximal interphalangeal joint and to further identify the ROI height. Finally, from the corrected image of a certain height, the ROI is obtained by using finger edges internal tangents as ROI boundaries. The results prove that our method is accurate and effective (EER rate of 0.12577), compared with other researchers (EER rate of 0.16296), and thus the proposed method for enhancing the performance of the system.

Extracting region of interest (ROI) is an important process for finger vein recognition system. It is determined the part of the image that we need for extracting features. This paper presents an ROI extraction method that overcomes the problems of finger rotation and displacement based on machine l...