Remote-Sensed Landsat TM Image Compression Using Classified Prediction and SPIHT with Low Complexity

摘要

A remote-sensed Landsat TM image compression method with low complexity is proposed using classified prediction and set partition in hierarchal trees (SPIHT). By classifying geographical regions with similar prediction coefficients, the proposed method can effectively remove spectral redundancy with a minimal level of complexity. In addition, a 3D SPIHT is performed using bands that are arranged based on the magnitude of the difference image value between the predicted and the original image. This method can be utilized for real-time processing because it is very simple and low in complexity. The proposed coder demon-strated a notably improved performance when compared with a conventional method used in remote-sensed TM image compression.