Researchers in the Statistics and Computer Science Departments at UCLA have developed a method for image compression that is 5x more efficient than JPEG image coding.
JPEG is a standard image compression algorithm as the method by which the digital image is stored and is commonly used in digital cameras. The compression process relies on a series of algorithms that approximate the color spectrum in predefined regions of pixels. Because the process is serial and is performed in the same manner across the whole image, the compression rate has no optimization routine and therefore suffers. Furthermore, the JPEG compression algorithm does not extract any features or structures that make it streamlined for further image processing if need be.
Researchers in the Statistics and Computer Science Departments at UCLA have developed a method for image compression that is 5x more efficient than JPEG image coding. They have circumvented the serial procedure of JPEG compression and developed a method for encoding images that result in a smaller file size. Their technique is based on the development of a ‘primal sketch’ which is extracted from an image based on the portion of the image with distinguishable elements. A library is generated from these elements which can then be applied to the entire image for compression. Additionally, since this new algorithm extracts features and structures, the resulting compressed image can be further analyzed by image processing techniques in later stages.
|United States Of America||Issued Patent||7,567,715||07/28/2009||2004-563|