A memory efficient architecture for compressed-domain texture feature extraction

Texture is a useful feature for representing the content of many real-world images. In order to reduce the time and memory overhead of traditional decode-process texture feature extraction systems, a compressed-domain approach is needed. This paper presents a new technique and hardware solution for obtaining accurate texture features from the compressed domain while using less memory and computational time. The architecture achieves these savings by reducing and simplifying the calculations performed. The design uses simple modules and can easily be extended to include a wide variety of image sizes and bit rates.