Selective removal of cr6 From water by biological waste materials Derived magnetic nanoadsorbents

Abstract

Face recognition framework is an innovation that is equipped newlineto recognize an individual from a computerized picture or video outline newlinefrom a video source. One of the significant technologies in this newlinedemanding technical environment is an Artificial Intelligence. Face newlinerecognition may also find applications in various wide varieties of fields newlinethat include recognizing and tracking of criminal in law and newlineenforcement field, surveillance through closed circuit television and for newlinesecurity purposes etc. Even though, there are many advantages of face newlinerecognition but the research is still going on to solve the face newlinerecognition issues under the unconstrained environmental conditions newlinesuch as poor lighting, posture variation, blurring and occlusions. Hence, newlinehybrid algorithms are developed and analysed in this research work newlineagainst the unconstrained images. The research work is carried out in newlinethree modules. Each module is explaining about the hybrid algorithm newlinethat is developed for handling the affected images in an efficient newlinemanner. In the initial stage of this research work, Directional Binary newlineCode (DBC) is preferred to extract the features from all orientations to newlinereduce the dimension and then Eigen values are calculated for the newlineextracted weighted sum of vectors to detect face and non-face regions newlineand finally it is trained and tested with Convolutional neural networks newlinefor extracting essential features and classification that improves the newlinerecognition accuracy and reduces the time for computation with newlinedatabases having minimal faces which means this algorithm works well newlineover the minimum number of samples. Then another module Hybrid newlineRPSM_CNN is proposed to recognize the face from the partially newlinevi newlineavailable features by using Robust point set matching that will extract newlinethe key points for matching and recognizing the image when applied newlineover the min-max classifier layer and then the generated feature vectors newlineare propagated to the further layers of CNN for classification and newlinerepresentation which in turn reduces the complexity of the CN