The goal is to reconstruct the image using the top singular values such that the trade off between the error of and image quality is optimum. This helps in also compressing the image and also shows the error plots and norms of the diffecrent channels. Here the image is broken down to 3 channels and then reconstruted from their individual singular values.