STUDY AND ANALYSIS OF IMAGE DECOLORIZATION TECHNIQUES

Abstract

Color information is often redundant when it comes to recognizing critical edges and features in a wide range of use cases. In image processing, converting a color image to grayscale helps eliminate redundant or non-essential color information. The current proposal is a novel approach for transforming a RGB image into Grayscale image based on Fusion of ICA and PCA. ICA is applied on color image to find independent components to make unique structures clear and the ICA output is 3 components per pixel. Applying PCA on the ICA output to find linear combination (principal component) which captures maximum overall variation and selecting first Principal component (PC1) which is most information rich and normalizing the PCA output. For Better contrast and clarity at the local level, the resultant grayscale image is processed with Contrast Limited Adaptive Histogram Equalization. This approach yields a grayscale image that better reflects the image's intrinsic intensity distribution while preserving more perceptual color information than traditional methods. Qualitative analysis has been done, and it shows that proposed method outperforms existing state-of-art techniques.