Elham Askari - Ph.D. Student of Computer Engineering, Science and Research Branch Islamic Azad University
Elham Ghasemi - Assistant Professor, Science and Research Branch Islamic Azad University
Ali Broumandnia - Ph.D. Student of Computer Engineering, Science and Research Branch Islamic Azad University

Image segmentation is an important research area in computer vision and many image segmentation methods have been proposed, therefore it is necessary to be able to evaluate the performance of image segmentation algorithms objectively. To date, the most common method for evaluating the effectiveness of a segmentation method is supervised, in which a segmented image is compared quantitatively against a manually segmented image. The evaluation methods that require user assistance are impractical in many vision applications and decrease the depth of evaluation, so unsupervised methods have been proposed. This paper have been presented a new unsupervised metric to evaluate the accuracy of IR image segmentation algorithms based on difference gray value of each pixel from mean gray value density of its region. We enumerate some suitable segmentation algorithms for IR images and then we evaluated them. Experimental results were obtained for a selection of IR images from OTCBVS Data Set and demonstrated that our metric is a proper measure for comparing IR image segmentation algorithms.

unsupervised metric; IR image; objective evaluation; image segmentation