SPATIOTEMPORAL PATTERN-RECOGNITION USING HIDDEN MARKOV-MODELS

A spatio-temporal method for identifying objects contained in an image sequence is presented The Hidden Markov Model (HMM) technique is used as the classification algorithm, making classification decisions base d on a spatio-temporal sequence of observed object features. A five cl ass problem is considered Classification accuracies of 100% and 99.7% are obtained for sequences of images generated over two separate regio ns of viewing positions. HMMs trained on image sequences of the object s moving in opposite directions showed a 98.1% successful classificati on rate by class and direction of movement. The HMM technique proved r obust to image corruption with additive correlated noise and had a hig her accuracy than a single-look nearest neighbor method. A real image sequence of one of the objects used was successfully recognized with t he HMMs trained on synthetic data. This study shows the temporal chang es that observed feature vectors undergo due to object motion hold inf ormation that can yield superior classification accuracy when compared with single-frame techniques.