A new computational model based face recognition system with edge extraction scheme is presented in this research. The proposed model has been built centering on some simple point spread operators, which are easily constructed from a set of orthogonal polynomials. One specialty of these point-spread operators is that they can be used in transforming vis-a-vis approximating 2D monochrome image regions. Also a complete set of difference operators are configured from these point-spread operators. Initially, we detect the face from the given input image using an edge extraction scheme, derived as maximizing the signal to noise ratio due to operator’s response supported by the proposed orthogonal polynomials. Simple procedures are derived to compute characteristic subsets of coefficients of the proposed transformation that represent important features, are considered for face recognition on the face detected input image. The proposed face recognition system is tested with The Yale database and also compared with Discrete Cosine Transform based face recognition system, Principle Component Analysis based face recognition system and fisher face recognition system.