A new algorithm for detection of lung boundary

A solution to the image segmentation problem, especially boundary detection for chest X-ray images, is presented. The properties of the boundary are determined by rescaling and thresholding of the gray-level image, Laplacian filtering, and a contour-following algorithm. A heuristic lung boundary searching algorithm is then used to make a sharp distinction between the lung region and the exterior of the lung. Some experimental results that show the fast and accurate performance of this search algorithm in comparison with others are given. >