Blur Identification with Image Restoration based on Application of Data Mining Techniques

Abstract

Image mining deals with extraction of implicit knowledge from the images. The blur is the most important degradation present in the digital images. Conventionally, the blur in the image is identified only with prior information of the Point Spread Function (PSF). If PSF of the image is unknown, the blur cannot be identified and restored using the conventional methods. This paper proposes soft computing based blur identification with image restoration for unknown PSF thereby achieving very high classification accuracy with minimum computational time. The proposed work consists of different stages such as phase preserving image de-noising, Blur identification using Multi-Layer neural network based on Multi-Valued Neurons (MLMVN), Image restoration using adaptive de-convolution technique, Biogeography-Based Optimization algorithm (BBO) for optimizing the neural network parameters. The hybrid version of MLMVN with BBO is proposed to search global minimum of the error performance surface of a blind image restoration problem that improves the computational speed and accuracy.