Research On Image Compression Algorithm Based On Classification

With the fast development of multimedia technology and data communication technology, more and more digital image data need to be stored and transported. Because storage capacity and transmission bandwidth are always very limited, the image compression technology has become a key to resolving the problem of storage and transportation and it is even more so for satellite remote sensing. Therefore, researching more effective compression algorithms is of significant meanings of theories and practices.At present, various compression algorithms are endless, most of the research work in image compression field focus on researching and improving the algorithms, while the researches about compression combining image classification are seldom be concerned with. The images with different characteristics can get different compression results. In order to make the image characteristics and compression algorithm better match, and improve adaptive capacity and compression quality of algorithms, a new algorithm of compression based on image classification is presented in this thesis. Through the performance simulation and comparison of existing image compression algorithms (JPEG and JPEG2000), the algorithm takes image statistical properties and grads which characterizes image edges into consideration, extracts grads which be sensitive to Algorithm performance evaluation and image quality as the classification feature, and finally establishes a new kind of image classification according to the selection of optimal algorithm of adaptive image compression algorithm model. In the image compression procedure, we can choose different compression methods and determine related parameters according to the size of the average gradient or image classification.A large number of experimental data shows that the new image compression algorithm has good compression effect. Compared with existing algorithms which directly compress images without classification, the new image compression algorithm improves the compression performance efficiently. From a macro point of view, the algorithm can significantly improve performances of a number of images, and from the micro point of view, to the image which is partitioned by the various images, a good result can also be obtained using this Classification compression algorithm.In a summary, with an objective of image compression, a classification processing about image using image grads in a certain sense is proposed, and is introduced to the field of image compression. The expected results are achieved also.