Indexing spatial content of still texture in MPEG-4

There has been a rapid increase in audio-visual information in digital form with the ever-increasing popularity of the Internet. The Moving Pictures Expert Group has recently presented a video compression and content manipulation standard, called MPEG-4 [1]. This was developed in response to the growing need for a coding method that can facilitate access to visual objects in natural and synthetic moving pictures in a wide variety of applications. In this paper, we propose a method for indexing the spatial content of Still Texture in MPEG-4 Video. There has been a number of approaches proposed recently [1],[3],[5],[6], which address content based indexing using features such as, color, shape and/or texture. We note that the still texture information is encoded using Wavelet Transform in MPEG-4. Since still texture can have an arbitrary shape, we propose to employ a combination of shape, color and texture features. The shape features are derived from the alpha plane and represented using Cubic B-spline approximation. The color and texture features are extracted directly from the wavelet compressed domain. An important attribute of the proposed approach is also the introduction of the novel concept of “containment” of spatial features for content based indexing.