Optimizing the Lz77 Dictionary Algorithm for Lossless Image Encoding

The article describes three modifications of the LZ77 dictionary algorithm for lossless image compression in the process of sequential bypass pixels: storing offsets to identical sequences in pixels, not in individual components; performing a search for identical sequences first, starting from adjacent previously processed pixels; performing a search for identical sequences not only in the horizontal but also in the vertical direction. The first of these modifications is shown to improve compression by using smaller values for storing offsets and increasing the dictionary of the LZ77 algorithm threefold. The second modification forms small offsets to adjacent pixels of the previous row. And the third one finds longer identical sequences. Storing offsets in pixels and separately searching for identical sequences from adjacent processed pixels is recommended to use in graphic formats, as they improve compression with almost no impact on encoding and decoding time. The search for identical sequences in two directions is suggested to be used only in archives, because the implementation of this modification slows down both encoding and decoding, improving the compression of only individual images. On the well-known ACT test set, it is shown that the application of the proposed modifications together with the simultaneous search of the same sequences in three dictionaries makes it possible to reduce image compression coefficients by an average of 0.18 bpb.