We present here a new statistical lossless data compression technique. Its performance is a bit higher than the one of arithmetic coding. It can be used in combination with other data transforms to create powerful compressing applications. According to our knowledge the principle, on which the method is built, is new. It is therefore likely that it will be an interesting topic for further research, testing, and improvements. The presented method can be often used as a replacement of some existing methods (Huffman and arithmetic coding) as it often gives better results. This method gives interesting results if applied as final phase of Burrows-Wheeler Transform based compression. We suppose that it can be very useful in compression of document databases. It can be also used for image compression, for example as a step processing the output of dicrete cosine transformation (after quantization) used in JPEG. The hardware support can highly benefit from the fact that this method cen be efficiently pipelined what promises high throughput.
展开▼
