Contemporary digital image processing applications require a suitable tool for further applications and processing. In this paper, a Singular Value Decomposition (SVD) based new sub-band decomposition and multi-resolution representation of digital colour images is proposed. Simulation was performed using MATLAB on Lenna image and the resultant sub-bands, which contained different directional details, are shown. Through a quantitative analysis, it is justified that the proposed method is better than other contemporary methods as it is suitable for compression and other image processing applications.