Image compression is a process of reducing the number of bits needed to represent an image. The goal is to optimize storage spaces, facilitate their transmission through the network and thus promote telemedicine. Over the years, several compression algorithms have distinguished themselves by their ability to reduce the size of the image while maintaining an acceptable visual appearance. These include the JPEG standard, the JPEG2000 standard and many others. The principle of these algorithms is essentially based on the reduction wavelet coefficients according to the singularity of the image. In this article, a new approach is proposed. The goal of this approach is to zero the wavelet coefficients regardless of the singularity of the image. To achieve this goal, our algorithm segments into three fundamental parts. The first part consists in breaking down the image into sub-bands through the QWT formalism. Subsequently, in order to obtain orthogonal matrices, we break down the matrices of the recently obtained sub-bands into singular values. The objective of these matrices is to exploit the redundancy present in the image while putting most wavelet coefficients to zero without, however significantly degrading the visual aspect of the image. To close the algorithm, we apply a thresholding function to the previously obtained wavelet coefficients. The method was evaluated by computer performance criteria such asand by human visual system performance criteria such as. These criteria are used to judge the quality of the reconstructed image and the compression ratio.