Abstract:
In order to extract the complete palmprint principal line which is close to the natural growth regulation, a method of palmprint principal line extraction based on multi-direction filtering and neighborhood de-noising is proposed. This method firstly preprocesses the whole palm image to obtain the ROI (Region of Interest) palmprint map with uniform size in gray-scale. Then it filters the ROI image with 4 directional detectors, and performs bottom cap operation on the filtered results to obtain the rough extraction results. Finally, binary processing and neighborhood de-noising are carried out on the rough principal line. Comparing the proposed method with the state-of-art methods in an open data set, the subjective judgment of the main line image extracted by the proposed method is more continuous and its objective PSNR value is about 2.3153 dB higher than the mean value of the compared methods. It shows that the proposed method has better effect and can effectively remove the broken line and isolated point noise so as to extract the natural principal line of palm print completely and effectively.