A Study on Multi-Focus Image Fusion Based on NSCT and Focused Area Detection

In this paper, a study on multi-focus image fusion methods having the ability to conquer the problems in transform domain method and to avoid the disabilities occurring in spatial domain method has been introduced. To gain an image with high quality and with all spectral and spatial information, it is important to recover the images from various multi-sensor images. Image fusions are mainly classified as transform domain image fusion and spatial domain image fusion. The main aim in multi-focus image fusion is to overcome the limitation in camera depth. The first step is to decompose the images using NSCT thus dividing the images into low and high frequency components. Then applying certain fusion rules with respect to the low frequency and high frequency an initial fusion image is obtained. By comparing the similarities between the initial fused images and source images a decision map is obtained by certain mathematical morphological procedures. The final image fusion is based on the decision map by selecting the pixels from the focused areas and obtaining the pixels at the boundaries as their respective pixels in the initial fused image. Keywords— Depth of field, fusion rules, human visual system, image fusion, mathematical morphology, spectral information.