In infrared thermographic testing, a background reflection is one of the main causes of a false detection. The authors developed an image processing program that can reduce the effect of the background reflection from a thermal image. The program mainly consists of two parts. The first part is the correction of emissivity which depends on a face angle. The emissivity is changed not only by a material property and a surface roughness but also by a face angle. Especially in a test for a high building and a floating roof tank, the thermal image includes a wide range of face angle. In order to solve this problem, by using the theoretical equation of emissivity, the angle correction of emissivity is carried out for each pixel in the infrared image. It is confirmed that the first part of this program can correct efficiently the emissivity change for a flat plate and a curved surface which have a wide range of face angle respectively. The second part of the program is the reduction of a background reflection. The emissivity change causes the reflectivity change and, as a result, this changes the effect of the background reflection. To reduce this effect, the background heat source is measured separately and then is subtracted from the infrared image. At this time, the specular reflectivity should be used. Consequently, this program can reduce the effect of reflection from the background heat source and extract the radiation of the flaw part from the complex thermal image.