Adaptive Segmentation and Boundary Detection for Images with Complex Lighting Environment

V.K. Asari (USA)


Adaptive thresholding, region growing, quadtree structure, Iris filter, boundary detection


A new technique for the automatic extraction of object region and boundary from images with complex lighting environment is presented in this paper. Segmentation procedure begins with the computation of an optimum threshold to distinguish the darker regions in the image. The image thresholding problem is formulated as an adaptive technique that progressively thresholds the histogram of an image. The thresholding algorithm uses the spatial distribution characteristics of an image and automates the thresholding process. The center of mass of this thresholded region acts as a seed for further processing. Then the exact object region is obtained by using an Iris filter, which operates in the gradient space of the image. A quad-structure based technique is used to enhance the speed of region search significantly. A boundary thinning and connecting algorithm based on the application of a novel search window on the preliminary boundary is also presented to obtain a connected single pixel width object boundary. The new method does not need a priori knowledge about the image characteristics. The performance of the new algorithm is evaluated by extracting the lumen region and boundary from gastrointestinal images. One of the main advantages of the new procedure is its high-speed response, which makes the real-time extraction of the lumen feasible.

Important Links:

Go Back