THE SOLUTION
	IMAGE COMPRESSION ALGORITHM SUMMARY
	Computer Science and Mathematics
	The RGB (red-green-blue) components of neighboring pixels of an image have a great deal of similarity. The degree of similarity between neighboring pixels is expressed by their autocorrelation function. It has been established that the value of the autocorrelation between neighboring pixels is very high, about 0.95 (95%). Since neighboring pixels carry similar information, the idea is to find an algorithm that keeps only the uncommon information between neighboring pixels, but retains the common information only once, thus avoiding repetition. This process is known as decorrelation. Our method of decorrelation belongs in the family of subband algorithms, where we divide the image into subbands of different frequencies. Only one of the subbands has the maximum information related to the image and for that one we have devised efficient data structure algorithms to store the information. Other subbands carry some information that is needed in order to obtain high image quality, but most of the information can be compressed at a high rate, without significant loss. Our contribution is having a new way to construct the subbands and a new way to determine which information is important and our own efficient datastructures for storing the information. This combination allows us to achieve high compression ratios for true color still photography with high resolution, greater than the competition. The current software is new. We have implemented a version of our algorithm that has a current compression ratio which exceeds 100 to 1 for graphic images, with virtually lossless resolution. The mathematics indicate that ratios will exceed 300 to 400 to 1 and can be achieved during the ongoing future development of the technology which is presently in Phase 1.