Recognition system for automatic diagnosis of knee ligaments

A recognition system for diagnosis of knee ligaments is described. Information about the knee is obtained by holo-interferometry. Fringe patterns on the reconstructed image are analyzed using image processing and pattern recognition techniques. Methods and algorithms for segmenting fringe area, extracting knee features, and classifying abnormal knees have been developed and tested. Twenty-five holo-interferograms have been inspected using the system. Automatic recognition of abnormal knees has been, realized. The degree of correct recognition is 84%. Only macro features of the original fringes are used in the analysis