Target segmentation using a plane with total least square error in nonhomogeneous infrared images

Abstract

In this paper, we propose a novel method to effectively segment a target regions in an ROI(Region of Interest) detected from a non-homogeneous infrared image. The proposed method is composed of two steps. In the first step, we apply a TLS(Total Least Square) technique based on PCA(Principal Component Analysis) to an IR image. Consequently, a plane vector and a SSE(Sum of the Squared Error) value is generated for estimate the non-homogeneous environment. In the final step, an angle between the brightness plane and a plane vector is calculated. Then, pixels with upper values than the plane and labeling numbers are updated. Consequently, in order to exactly segment a target region, three conditions of SSE, angle and labeling values are iteratively updated. Experimental results show that the proposed method achieves better performance of segmenting targets under cluttered background environment than the existing method.