Independent component analysis for spatial object recognition with applications of information theory synthesis

Each moving object contains particular unique signatures that can be used for pattern classification via object recognition and identification. Information extracted from the spatial object feature recognition can be provided by independent basis functions to represent actual physical attributes of the moving objects. Compared with principal component analysis, independent component analysis is a special feature extraction approach for blind signal separation, where an object is labeled to a special class. Some underlying factors or sources can be captured in a statistical sense. The true color image is composed of red, green and blue components which are perpendicular to each other. These components may serve as a basis to be synthesized using independent component analysis. Each individual signature indicates unique information that can be evaluated using information theory. Thus, the quantitative measures of the color component energy, discrete entropy and relative entropy have been introduced to independent component analysis issues for recognition of moving objects.