| dc.description.abstract | Palm-print trait based biometric identification has emerged as a most powerful tool to recognize a person’s identity. It is used in commercial and forensic applications. In common, it considers high 400 dpi (Dots per Inch) or more is high resolution and 150 dpi or less is low resolution. Earlier research projects have been showed that high-resolution palm images are capable to extract ridges, singular points and minutia points as features. Low-resolution images have capability to extract principal lines, wrinkles, and texture. Therefore, researches which are based on palm prints primarily focus on highresolution palm images. Therefore the main purpose of this research was to provide a Fast and Accuracy Palmprint Recognition method for low-quality images using image processing with feature extraction. In this research, the database which was used for this is mainly consisted with the palm images which were collected from students. In feature extraction process, low pass filter was used to remove the noises of images. Elongated and tubular structures were enhanced in the noise removed images to highlight major lines by using Hessian-based multiscale filtering. Segmentation process in the original image transforms the original image in to a binary image in which ridges area fully colored in one tone and the background in the opposite tone. Threshold binary image was applied to some morphological operations to extracting the better results. The palm-print matching process which is also called as the template matching was mainly based on the normalized cross correlation in Fourier domain (phase correlation). This process was done by pixel-by-pixel basis. Results of this process have shown a higher Genuine Acceptance Rate for lower False Acceptance Rate and False Reject Rate. | en_US |
