Image compression, F-transform, Image similarity

The aim of the work is to continue in improvement of the image compression algorithm based on the F-transform. The image is decomposed into blocks and characterized by the F-transform components. The latter constitutes a simple lossy compression. For better quality of reconstructed images, we compress certain areas (neighborhoods of edges) nonlossy. The proposed approach is based on establishing similarity between various blocks and making compression of only one representative. Last but not least, the proposed compression algorithm is supplied with smart technique of joining adjacent blocks.