A SYSTEM FOR INDIVIDUAL CARE IN PERVASIVE ENVIRONMENT

摘要

Bio-metric is a technology which is evolving very rapidly and can be applied in number ofapplications like criminal identification; security etc. There are number of systems whichrequires authentication in order to determine the individual’s identity. Main role of theseschemes is to allow data to a valid user not fake one. It is used to identify the identity of a personby using some methods and there are number of methods for this technology like facerecognition, voice recognition, iris recognition, palm recognition, fingerprint recognition etc.Biometric systems are automated systems for identifying user through physiological orbehavioural characteristics. Physiological are related to the shape of the body. The oldest traitsthat have been used for more than 100 years are fingerprints. Other examples are facerecognition, hand geometry and iris recognition. Behavioural are related to the behaviour of aperson. The first characteristic to be used is the signature. More modern approaches are the studyof keystroke dynamics and of voice recognition. Trust based secure environment for anindividual alone at home is provided in this work. This work includes a combination of twomodified biometric techniques face recognition and voice recognition. Enhancement andimprovement of face and voice recognition system is done by integrating algorithms likePrinciple Component Analysis and Independent Component Analysis for face recognition andMel Frequency Cepstral Coefficient with RASTA- Perceptual Linear Prediction for VoiceRecognition. Combination of these algorithms is used for extraction of features and forrecognition of face and voice, Feed Forward Neural Network is used as a classifier. We combineimproved Face Recognition with improved version of Voice Recognition approach to providemore secure environment to an individual. Combination of Voice and Face Recognition is used ifwe want to identify a person. If Face is recognised of a person, then go to Voice Recognition, ifboth are recognised accurately of an individual then access will be granted otherwise if any oneparameter is not recognised properly in that case we will calculate the similarity index of thatindividual, if it shows the value greater than equal to 75% then only permission will be grantedelse the person will be declared as unknown. This combined approach of Face Recognition withVoice Recognition performs better than previous approaches and the experiments we performgive results with 100% accuracy by using number of images and audio files as a database. All theinformation will be send to cell phone of family members by using MATLAB tool and Dongle.In addition to this, an application is also used side by side i.e. fingertip application whichcontinuously monitors individual’s health parameters like bp, heart rate etc. by using sensorbased camera on smart phone and matlab tool. By following the same process, send message tofamily members if value of parameters is not in range.