Handwritten Signature Verification using Statistical Functions for Signature Image

This research suggests a new method that aims to verify the manual signature image which is written by person, and specify whether this signature back to this person or that forged signature. This was done by extracting geometric features of the signature image and applying statistical functions on them as a way to verify the signature of that person. The features from the signature image have been extracted on many stages so a signature image has been transformed from the gray scale to binary format, and then extracting the statistical features from the original signature image which is the maximum value from the most repeated values in the ones' coordination line that determine the signature shape, in addition to the number of ones which also determine the signature shape. Finally two ranges have been identified for the values accepted for original signature image. By the same way, statistical features have been extracted from the foreign signature image and tested if they aggregate within the specified domain of acceptable values. This research also includes the results of the proposed approach that compared with the previous methods in this scope. The proposed method has been tested to the data base consisting of 16200 signatures back to 300 persons, and as a result the signature image has been verified with a good percentage.

Design a Weighting and Fusion Model of Iris Features To Improve Performance Of Iris Recognition Systems

The study suggests designing a weighting model for iris features and selection of the best ones to show the effect of weighting and selection process on system performance. The search introduces a new weighting and fusion algorithm depends on the inter and intra class differences and the fuzzy logic. The output of the algorithm is the feature’s weight of the selected features. The designed system consists of four stages which are iris segmentation, feature extraction, feature weighting_selection_fusion model implementation and recognition. System suggests using region descriptors for defining the center and radius of iris region, then the iris is cropped and transformed into the polar coordinates via rotation and selection of radius-size pixels of fixed window from center to circumference. Feature extraction stage is done by wavelet vertical details and the statistical metrics of 1st and 2nd derivative of normalized iris image. At weighting and fusion step the best features are selected and fused for classification stage which is done by distance classifier. The algorithm is applied on CASIA database which consists of iris images related to 250 persons. It achieved 100% segmentation precision and 98.7% recognition rate. The results show that segmentation algorithm is robust against illumination and rotation variations and occlusion by eye lash and lid, and the weighting_selection_fusion algorithm enhances the system performance.

Personal Identification via Handprint with use of Zero-Crossing Technology

Personal identification based on handprint has been gaining more attention with the increasing needs of high level of security. In this study a novel approach for human recognition based on handprint is proposed. Wavelet transform was used to extract features presented in the palm image based on wavelet zero-crossing method. Firstly the wavelet transform of the whole palm image at the fourth level was worked out, which results in four matrices; three of them are detail matrices (i.e., horizontal, vertical and diagonal) as well as one approximation matrix. Throughout this study, only the detail matrices were used because the required information (i.e., hand lines and curves) is included in those matrices. Sixteen features were extracted from each detail matrix, and then arranged in one vector. Consequently, for each palm sample a feature vector consisting of 48 input features of the used neural network was obtained. For this purpose, a database consisting of 400 palm images belonging to 40 people at the rate of 10 images per person was built. Practical tests outcome showed that the designed system successfully indentified 91.36% of the tested images.