基于 NSCT 的自适应多判决图像融合方法

摘要

As the traditional fusion methods can not effective retain the target information of infrared image and the a-bundant background information of visible image,the adaptive and multi-judgement image fusion method for infrared and visible images based on NSCT is proposed.Firstly,the registered infrared and visible images from the same scene are decomposed by NSCT transform in multi-scale and multi-direction.By taking different fusion rules for the low and high frequency coefficients respectively,the low frequency coefficient is fused by regulatory factor and adaptive weighted aver-age based on regional energy,while the high frequency coefficient are processed by different fusion rules,that is to say, it is fused by using absolute maximum for high-level frequency coefficient,and it is fused by using multi-judgment fusion method based on contrast ratio for low-level frequency coefficient.Finally,the fusion image is obtained by performing in-verse NSCT transform.The experimental results indicate that by using subjective and objective evaluation,the proposed fusion method is better than the other image fusion method based on multi-scale analysis,and it can obtain more ideal fusion image.%针对传统的融合方法对于红外图像目标信息和可见光图像丰富的背景信息无法有效地保留到融合图像的问题，提出一种基于 NSCT 变换的自适应多判决红外与可见光图像融合方法。首先，将来自同一场景配准后的红外与可见光图像进行 NSCT 多尺度、多方向分解；然后，分别对高低频系数采取不同的融合规则：低频系数采用基于区域能量调节因子自适应加权的方法，对高频系数采用混合的融合方法，即对于高层，采用像素点绝对值选大的融合方法，对于低层采用基于对比度的多判决融合方法；最后，将得到的高低频融合系数进行逆 NSCT 变换得到融合图像。实验结果表明，经主客观评价，本文提出的融合方法优于文中其他基于多尺度分析的图像融合方法，可得到更理想的融合图像。