A novel algorithm for infrared image contrast enhancement based on neutrosophic sets

Abstract

ABSTRACT Due to the narrow thermal window of infrared (IR) image sensor array, the represented image becomes dim and its details become blurred. Different contrast and detail enhancement technologies were deployed to improve image quality. An enhancement algorithm with bilateral filter is a state of art method, which first transforms an infrared image into a base part and a detail part, and then expands the detail part and suppresses the base part to enhancement the contrast of the IR image. However, this method cannot efficiently distinguish the background and objectives of the detail part. As a result, the noise information gets amplified when the detail part is expanded, leading to increase noise and to obstruct the sharpening of the detail part. To solve these problems, a novel enhancement algorithm based on the neutrosophic sets is proposed, which transforms the detail part into the neutrosophic domain. Our method utilises three membership functions of pixels in the neutrosophic domain: T (True), F (False) and I (Indeterminacy), which correspond to objective, background and transitional regions of image, respectively. The proposed algorithm is verified and compared with other existing algorithms. The experiment results show that the proposed algorithm can effectively enhance the contrast and preserve the details of an infrared image.