Dominant Feature Prediction By Improved Structural Similarity Computation
K. Indra Gandhi, Sugitha Janarthanan, Rakshith Sathish, A. Surendar · 2020
Recently features based on Machine learning approaches, such as face detection model trained with Convolutional Architecture for Fast Feature Embedding (caffe) framework, have been shown to be more effective than conventional hand crafted features and have become the state-of-the-art in object detection, including face detection. Detected Faces can be recognised using Support Vector Machine (SVM) and Convolutional Neural Network (CNN) which were widely used in monitoring systems and other face recognition applications. In this work, the features that are common within family members are detected by computing the threshold dynamically. Structural Similarity Index (SSIM) is a method for comparing two images with the perceptual metric. Using SSIM we can compare the extracted features to predict the dominant feature among the family members. Certain features that are similar between persons cannot be determined even through naked eyes. Hence this application comes into play that could determine the similar and dominant feature with adequate accuracy from the detected and recognized face. Age progression is most often used as a forensics tool by law enforcement. Using the inherited features extracted from the family members helps in predicting age progression of an individual from that family. By dynamically computing the threshold, the dominant feature of a family has been computed which improves the precision of the system. To reach maximum accuracy in age progression dominant features along with inherited features extracted from the image are used in combination with existing age progression technique. For the high performance of the system, dynamic variation technique in fixing the threshold and feature extraction have been developed.