Determining Correspondence in Stereovision Images of Patients with Faulty Posture

In the paper we present an algorithm for finding the correspondence between stereo pairs acquired during diagnosing faulty posture using the photogram-metric method. The coordinates of markers from images constitute the vertices of graphs, between which isomorphism is searched for. The paper proposes a method for finding graph correspondence and for determining similarity between graphs. The precision of the algorithm was verified for various patients' positions of patients as well as for dummy graph vertices, generated for test purposes. The developed algorithm is employed in clinical research, as an automatic system for faulty posture diagnosing using two stereovision images.