<![CDATA[This study examines the application of the Singular Value Decomposition (SVD) method as a digital image compression technique on the Yogyakarta Cosmological Axis object which is used as a digital learning medium in vocational education. The background of this study is based on the need for high-quality visual media with efficient file sizes for easy storage, transmission, and access through digital-based learning systems. The study uses an experimental quantitative approach with data in the form of high-resolution digital images processed through SVD-based compression stages. The research procedure includes image transformation into matrix form, matrix decomposition using SVD, selection of a number of dominant singular values (ranks), and reconstruction of the compressed image. The research data were analyzed using image quality evaluation parameters, namely Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Compression Ratio (CR). The results show that an increase in the rank value is directly proportional to an increase in the quality of the reconstructed image, as indicated by a decrease in the MSE value and an increase in the PSNR and SSIM values. Conversely, a decrease in the rank value results in a higher compression rate but is followed by a degradation in the visual quality of the image. Experimental data also shows that most of the visual information of an image can be represented by a small number of principal singular values, thus allowing for significant file size reduction without losing the important visual structure of the image object. Visually, the compressed image at a medium rank value is still considered suitable for use as a learning medium because the main details, object contours, and visual characteristics of the Yogyakarta Cosmological Axis can still be recognized well. These findings prove that the SVD method is effective as a mathematical-based image compression technique to support the development of efficient, informative, and contextual digital learning media based on local wisdom in vocational education]]>
Copyrights © 2026
