{
"title": "A Complexity-Based Approach in Image Compression using Neural Networks",
"authors": "Hadi Veisi, Mansour Jamzad",
"country": null,
"institution": null,
"volume": "35",
"journal": "International Journal of Computer, Electrical, Automation, Control and Information Engineering",
"pagesStart": 2619,
"pagesEnd": 2630,
"ISSN": "1307-6892",
"URL": "http:\/\/waset.org\/publications\/4800",
"abstract": "In this paper we present an adaptive method for image\r\ncompression that is based on complexity level of the image. The\r\nbasic compressor\/de-compressor structure of this method is a multilayer\r\nperceptron artificial neural network. In adaptive approach\r\ndifferent Back-Propagation artificial neural networks are used as\r\ncompressor and de-compressor and this is done by dividing the\r\nimage into blocks, computing the complexity of each block and then\r\nselecting one network for each block according to its complexity\r\nvalue. Three complexity measure methods, called Entropy, Activity\r\nand Pattern-based are used to determine the level of complexity in\r\nimage blocks and their ability in complexity estimation are evaluated\r\nand compared. In training and evaluation, each image block is\r\nassigned to a network based on its complexity value. Best-SNR is\r\nanother alternative in selecting compressor network for image blocks\r\nin evolution phase which chooses one of the trained networks such\r\nthat results best SNR in compressing the input image block. In our\r\nevaluations, best results are obtained when overlapping the blocks is\r\nallowed and choosing the networks in compressor is based on the\r\nBest-SNR. In this case, the results demonstrate superiority of this\r\nmethod comparing with previous similar works and JPEG standard\r\ncoding.",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 35, 2009"
}