Results (
English) 1:
[Copy]Copied!
In the development process of Offline Signature Recognition techniques has been used a lot in order to give Recognition for high performance and require effective Preprocessing shall encourage. The Feature Extraction process. High efficiency as well. For this edition of the above mentioned research, process development and performance with the Preprocessing Neural Network Pattern, which is the process that is using the technique of Preprocessing to reduce the variance of the signature with the concept that the variance of the performance information of the ta and the variance of Recognition appropriate information to.TA, high performance Java knowledge this technique is to create original images with greater density by increasing the thickness of the lines and the use of variation Coefficient (CV) is a measure of the Standard Deviation by the Mean thought that the trial process image data, with a value equal to the CV.Then go to properties with farming process Feature Extraction of images with signature of a Histogram of Gradient (HOG) and Histogram of Gradient Pyramid (PHOG), the result will be a Panama in artificial neural network. Continue to measure performance. The test will be used for data, photos, personal signature, there were 150 people were all 5739. The trial results will set performance from 91.75-100% for HOG and 92.1-100% for CV = 1 value to PHOG, optimizing for performance equal to 93.72. CV-HOG-100%
Being translated, please wait..
