Comparative Evaluation of Image Compression Techniques for High-Resolution Orthophoto Imagery

Abstract

The rapid growth of high-resolution data from UAV, LiDAR, and satellite platforms has significantly improved the accuracy and applicability of geospatial analysis, while simultaneously creating challenges related to large data volumes, storage, transmission, and computational efficiency. Image compression plays a critical role in addressing these challenges by reducing data size while maintaining acceptable visual and analytical quality. This study presents a comparative analysis of three widely used image compression techniques—JPEG2000 (lossless and lossy) and Enhanced Compression Wavelet (ECW)—applied to high-resolution LiDAR orthophotos. Compression performance was evaluated in terms of output size, compression ratio, and processing time, while image quality was assessed through both qualitative parameters (edge sharpness, spatial detail, color fidelity, radiometric consistency, texture, and geometric integrity) across multiple interpretation scales and quantitative approaches (MSSSIM, PSNR, ERGAS, and RASE metrics). Results indicate that JPEG2000 lossless preserves complete radiometric and spatial fidelity, making it ideal for high-precision analytical applications, although it offers limited compression efficiency. JPEG2000 lossy demonstrates the best overall balance, achieving high compression ratios with lower processing time while maintaining strong structural and spectral integrity, as evidenced by consistently high MSSSIM and PSNR values and low error metrics. In contrast, ECW achieves the highest compression ratios and smallest file sizes, making it highly efficient for storage and transmission; however, it exhibits greater variability, higher computational cost, and reduced performance in high-detail features. Overall, JPEG2000 lossy emerges as the most suitable compression method for general remote sensing applications, while JPEG2000 lossless is recommended for high-accuracy tasks and ECW for storage-efficient visualization and data dissemination.