Fusion of Infrared and Visible Image Based on Target Extraction and Contourlet Transform

Abstract

Listen

Translate article

The fusion of the infrared and visible images has been widely used in object recognition, night vision and military affairs, which is a popular research field. A new image fusion method based on target extraction and contourlet transform is proposed in this paper. Commonly, the traditional image fusion method neglect differences between the targets and background of the infrared and visible images, resulting in the poor distinct or weak identification of the fused image. In order to take full advantage of the differences, we extract firstly the interested targets of infrared image, which are fused with the visible image by the method of regional similarity. Therefore we obtain a new visible image with more target information, while reserve the visible background information. Secondly, to obtain more complementary information, contourlet transform is utilized to fuse the new visible image and the source infrared image. In addition, based on the different characters of low frequency and high frequency coefficients, we choose different rules to fuse the contourlet coefficients. In low frequency processing, the method based on the fuzzy theory is used, while we ascertain the high fusion coefficients by the Tenenbaum’s algorithm. Experiments are carried out and the results show that our method is effective and the fused images are better than those resulting from wavelet transform and contourlet transform both in visual quality and in quantitative evaluations.