A computer vision system for diagnosing scoliosis

Presents a computer vision system for diagnosing spinal deformations using moire images of the human back. The system consists of two major stages. A segmentation in which fringe contours are extracted using the infinite size symmetric exponential filter (ISEF) which is implemented in a parallel fashion. In the second stage, image interpretation is performed using some relevant features and the three dimensional surfaces that are reconstructed from the contour fringes by applying a parallel relaxation operator (Mota and Velasco operator for discrete relaxation). In order to have a real-time vision system, most of the system´s components are implemented in a parallel fashion. The segmentation stage uses both spatial and temporal parallelism for speeding up the fringe extraction. The proposed system has been tested on a set of noisy moire images, and the experimental results have shown its robustness and reliability for the recognition of most scoliosis deformities