Auto-associative neural network for sparse representation based face identification
Suprava Patnaik · 2017
Sparse representation is a novel methodology that has off late received substantial attention for image classification and recognition. This paper presents a PCA-based dictionary building for sparse recognition. Recursive least square based auto-associative neural network model has been used for principal component extraction. Suggested network structure supports data compression along with principal component extraction. The trained principal components are used as the atoms of the dictionary. Proposed approach performs the recognition task at a low dimension space, hence involves less computation. In the recognition phase, a given image is sparsely represented as a linear combination of neuronal weights, which corresponds to underlying principal components. Our extensive experimentation on publicly available face databases (lfw and GT) endorses that this approach is promising for real time applications and can further be applied for face recognition tasks.