Hybrid Image Segmentation & Energy minimization technique for the images with non-uniform light intensity

Image segmentation is a classical problem in computer vision and is of paramount importance to medical imaging. The segmentation is complicated by lack of clarity, the overlap of intensities and many other factors. We present a hybrid algorithm for obtaining segmentation of images that are subject to noise and multiplicative intensity nonuniformity. The algorithm is formulated by the combination of bias field correction algorithm and level set segmentation method. An adaptive local clustering criteria function can be integrated for the determination of the intensity nonuniformity pattern. This non-intensity pattern taken off from image using the conventional level set segmentation method. Energy minimization of the level set function enables us to simultaneously segment and correct the image. A MATLAB code has been implemented as per the proposed method and it gives us good result of segmentation in most of the cases. Our method simply increases the performance in terms of reduced complexity, reduced time and grater segmentation accuracy.