Abstract

JPEG-LS encoding uses a low complexity compression concept, taken from LOCO-I, to compress images without loss of resolution. The standard uses prediction based on pixels to the left, above, to the upper-left, and to the upper-right of the current pixel being decoded. The standard then uses Golomb-Rice coding or run-length coding to encode the prediction error.

The JPEG-LS standard has a much lower compression ratio than its lossy counterpart. This makes JPEG-LS undesirable for images immediately presented to the human eye, but very useful for images, such as those collected by satellite imagery, that are likely to be processed after transmission for extraction of otherwise imperceptible detail.

Because the amount of compression acheived is a function of the number of attemps at compressing the image, if an image will be downloaded many times, more effort should be initially expended compressing the image. This paper presents an equation describing the optimal relationship between compression effort and expected required bandwidth for JPEG-LS.