08-01-2014, 03:20 PM
SOM: A Computer Vision Technique for Face Sketch Featured Database
Abstract
Self-Organizing Maps (SOM’s) found to be an improved data management computer vision technique used for the closed matching of face vs sketch identification system based on neural network of untrained input images with trained database of images. Parameters for the SOM neural network are selected to be a minimum and maximum point for each row on a vector of training database. In this paper 64 minimum and 64 maximum pixel intensity values selected altogether using 8x8 image masking technique. Further for the design of SOM a set of 25 uniform image data used to create 5 different classes of a face image like left eye, right eye, nose, frontal face and lips for the training database. All the preprocessing for the image enhancement is done in the MATLAB software.
Abstract
Self-Organizing Maps (SOM’s) found to be an improved data management computer vision technique used for the closed matching of face vs sketch identification system based on neural network of untrained input images with trained database of images. Parameters for the SOM neural network are selected to be a minimum and maximum point for each row on a vector of training database. In this paper 64 minimum and 64 maximum pixel intensity values selected altogether using 8x8 image masking technique. Further for the design of SOM a set of 25 uniform image data used to create 5 different classes of a face image like left eye, right eye, nose, frontal face and lips for the training database. All the preprocessing for the image enhancement is done in the MATLAB software.