Application of Singular Value Decomposition technique for compressing images

Image processing is becoming increasingly important as imaging technology has advanced. A storage constraint might occur even when image quality is an influential factor. This means finding a way to reduce the volume of data while still retaining quality, since compactable systems and minimal space are more desirable in the current computing field. An image compression technique that is frequently used is singular value decomposition (SVD). SVD is a challenging and promising way to loosely compress images, given how many people use images now and how many different kinds of media there are. SVD can be employed to compress digital images by approximating the matrices that generate such images, thereby saving memory while quality is affected negligibly. The technique is a great tool for lowering image dimensions. However, SVD on a large dataset might be expensive and time-consuming. The current study focuses on its improvement and implements the proposed technique in a Python environment. We illustrate the concept of SVD, apply its technique to compress an image through the use of an improved SVD process, and further compare it with some existing techniques. The proposed technique was used to test and evaluate the compression of images under various r-terms, and the singular value characteristics were incorporated into image processing. By utilization of the proposed SVD technique, it was possible to compress a large image of dimension 4928 x 3264 pixels into a reduced 342 x 231 pixels with fair quality. The result has led to better image compression in terms of size, processing time, and errors.