Near-infrared Image Enhancement Method in IRFPA Based on Steerable Pyramid

摘要

Near-infrared imaging mainly uses near-infraredband ambient light imaging reflected by the target, which hasbetter atmospheric penetration performance and human skinpenetration performance than visible light imaging. Therefore,near-infrared imaging is widely used in military, medical andmany industrial production. Aiming at reducing the noise andimproving the contrast of the near-infrared images gained frominfrared focal plane array (IRFPA), the near-infrared imageenhancement method based on steerable pyramid is proposed inthis paper. First of all, the near-infrared image is decomposedinto multi-scales using the steerable pyramid model; then thecoefficients of low-frequency and high-frequency of the imageare obtained. In order to improve the contrast of the originalnear-infrared image, the coefficients with low-frequency arenonlinearly transformed through fuzzy-set theory. Then thecoefficients of high-frequency are dealt with threshold methodto reduce the noise. Next, these images are reconstructed. At last,anti-sharpening mask is used to highlight the details of the image.During the reconstruction, a adaptive interpolation algorithm isput forward to resolve the distortion problem in the steerablepyramid algorithm. The experimental results show that thisalgorithm has a good effect on the enhancement of near-infraredimages, and significantly improves the quality of near-infraredimage produced by IRFPA device. The comparison results withvarious algorithms show that our algorithm outperforms thestate-of-art in terms of contrast gain, comentropy, mean-squareerror, and peak signal to noise ratio. The experimental resultsillustrate that the proposed algorithm can efficiently enhancethe near-infrared image.