WSF-RBF based mining model to identify eye-Glasses worn people from face-images pool

Eye-Glasses or Spectacles are the wearing objects to gain vision correction. Wearing glasses can be identified as low vision problem, poor nighttime vision or partial eyesight. Some of the departments and organizations are sensitive to healthy eyesight while recruiting a person. To identify the eyeglasses wearing people from a pool of facial images is a challenge. In this paper, statistical feature formed data mining model is provided for categorizing the people with and without eyeglasses. In the early phase, the geometric notification is applied to localize the eye region. This extracted region is processed by phase variant Gaussian enabled moment feature analysis to generate the structural form of an image. Later on quantification is applied to the real eye region image and structured region under 16 different statistical features. This featured formed dataset is qualified under weight modeling based on feature type and strength. This weighed feature formation is learned for RBF(Radial Basis Function) to perform classification. The experimentation on dataset for LFW and FERET datasets. Multiple experiments are applied and comparative observations are taken against neural network and SVM methods applied directly to eye extracted images. The observations show that the framework improved the classification accuracy up to 19% for FERET dataset and up to 16% for LFW database. The weight adjustment in the framework subject to accuracy and data features.