Abstract: Aiming at the great difficulties such as infrared background modeling, a method of implanting real infrared target image into real infrared background image is proposed in this study. First of all, the gray scale of the target image and the background image are used to inverse the radiation brightness of the target, background around the target and the new background which is needed to implant the target. After that, using Stephen Boltzmann’s law, the blackbody equivalent temperature of target and background is calculated based on the radial brightness. According to the temperature consistency, target temperature under the new background is calculated by the blackbody equivalent temperature, and then the gray scale of the target under the new background is obtained. Considering the mutual radiation of target edge and the surrounding background, target edge pixels and background pixels are merged through the process of convolution. In the process of convolution, Gaussian kernel matrix is adopted to blur the pixels between the target edge and the surrounding background. Experimental results show that the method can implant target effectively, with higher image resolution and better preservation of the texture information. Furthermore, the fusion method is based on temperature field consistency which is more reliable. This study is expected to complete the target implantation from a new angle.

Key words: target implantation, temperature field, fusion

摘要： 针对红外背景与目标建模难度较大等问题，提出了基于温度场一致性将真实红外目标图像移植到红外背景图像的方法。首先，根据目标图像与背景图像灰度反演目标与目标周围背景以及需植入目标的新背景的辐亮度，并根据辐亮度计算目标与背景的黑体等效温度。利用温度关系，计算出目标在新背景下的温度，进而求出目标在新背景下的像素灰度值。考虑到目标边缘与周围环境的相互辐射影响，通过卷积过程融合目标边缘与背景像素点。采用高斯核矩阵来模糊目标与背景边缘像素。相关实验结果证明，该方法可有效移植目标，移植后的图像清晰度较高，纹理信息保留程度较好，而且是基于温度场的融合方法，比较真实可靠。该研究有望从一个新的角度完成目标移植。

关键词: 目标移植, 温度场, 融合