Random network learning and image compression

Digital image compression serves a wide range of applications. Encoding an image into fewer bits can be useful in reducing the storage requirements in image archival systems, or in decreasing the bandwidth for image transmission for applications such as teleconferencing and HDTV. Although some applications (e.g. medical imaging) require lossless compression, image compression usually introduces some loss in the original image. Another issue is the speed of compression and/or decompression, especially in real-time applications, In this paper the authors use a learning random neural network to achieve fast lossy image compression for gray level images.<>