Blur estimation using polynomial curve fitting algorithm for image restoration using blind deconvolution

In most of image related application such as medical image, remote monitoring and photography good focused image selection is important task. The goal is to select the good focused image for further computer analysis. Discrete Cosine Transform (DCT) has low cost and requires low computational time. The DCT is known for its high energy compaction property, which requires less computation time than that of wavelet transform, so it can be useful for implementation in micro chip to save memory and power. Therefore in this paper focus measure is calculated for various size of blur using Bayes spectral entropy based on DCT coefficients of an image. Using curve fitting algorithm the blur is estimated from the focus measure values for image restoration using blind deconvolution.