Tracking of moving objects in scenery by subspace projection using independent component analysis

A system is developed for tracking moving objects through natural scenery. A technique is presented for performing change detection on imagery to determine the difference between two images or a sequence of images. From there an algorithm is presented to detect the presence of a moving objects. Then the development of a Variable Structure Interacting Multiple Model tracking filter is presented.; The method of performing change detection is based upon the concept of image subspace projection. A set of “basis” image maps are formed and when combined with a mixing matrix can recreate the original image. The subsequent images are then projected onto the base image. The projected image is then subtracted from the original image to perform the change detection. The basis images are computed by applying Independent Component Analysis (ICA) to the sampled image. The ICA algorithm used was the fast fixed-point algorithm developed by Aapo Havarien and Erkki Oja. A spatial filter and an adaptive threshold filter are applied next. The resulting locations of changes in the images are passed to the tracking filter. Tracking is performed through the use of multiple motion models. The filter's kinematic motion models are adaptive added or deleted as required by the moving object's dynamics. The moving object's state is estimated through extended Kalman filtering.