In the image collection of data the volume is ever increasing in the various sectors of medicines, science, security as well as other fields that bought the importance of extracting knowledge’s  The computer vision has the main challenge of face recognition or classification. This paper shows the details of the development of a real time face recognition system that is focused for operating in a less constrained area.

At the beginning it reviewed the popular techniques that are used for recognition of faces followed by the details of each step and explanation of the ideas behind it that lead to these techniques. It not only helps in the task of pattern recognition but the neural network process is a Spa application for the face recognition. In our study we have developed a face recognition system that is based on the step error tolerance back-propagation neural network.

Flexible and compact design id provided by the SET-BPN and it also help to reduce the step wise errors. This will make the system easy and readily operable. At the same time it will also provide the best results for classification. For system analysis we make several tests using the real data. According to the empirical results the method that is proposed to greatly enhance the speed of recognition of the the feature matching step.

In the service of security the face recognition plays the best role and in this document we proposed a model of recognition of the face using the step are tolerance back-propagation neural network concept and the processing of digital image which is simple, fast and accurate in constraint surroundings like in household or offices. This system helps in detection of human face and it also enables face recognition and eye localization in the speed almost close to the real time. There are many other advantages of this proposed method.