Identification Method of Inner Knuckle Print Based on Geometric Features

Previous studies have shown that the IKP (inner knuckle print) factors are unique and can be used for identity recognition. For the IKP features positioning is not accurate and hand information utilization is insufficient, this paper proposes a new identification method based on the geometric features of the IKP. First, the collected palm images are preprocessed, and the fingertips position are located by the improved center of gravity distance method. Next, the finger valley points and the reference points of the IKP are detected by using the gray gradient changes, and then the ROI (regions of interest) of the IKP are determined, Harris corner detection and K-means clustering algorithm are used to locate the centroid positions of each ROI, and 20 feature quantities for identification are constructed from the geometric features of the 12 centroids and the 4 fingertips. Finally, Knearest neighbor algorithm is used to calculate Euclidean distance of the features for feature matching. In order to verify the method proposed in this paper, a self-built small database is used for testing, and the experiment result show that the recognition rate is 98.39%, which verifies the feasibility and effectiveness of this method.