Fractures and damages can be resulted from physical stresses in the flawed parts of an aircraft. The exact detection of internal defects is possible by radiography testing. The determination of the defects locations by the experts depends on the accuracy, skill and quality of the radiographs. Many of the produced images by industrial radiography are not clear, and it is difficult to accurately detect their defects; hence, the processing methods can help to better investigation of the defects. These images suffer from some opaqueness due to the inherent dispersion of X-rays. In this study, a NON-LOCAL mean method based on the detection of similar pixels in a neighboring area is used to identify corrosion areas. In the method, the image is split into smaller windows and the similar pixel is found on these parts. The results of the method show that, due to the lack of noise function in radiographic images, the background elimination method is more suitable for this algorithm. In this research, the large amount of standard deviation of the noise is considered and background is extracted, the obtained image was reduced from the original radiographic image. The reconstructed image has sharp edges that more clearly indicates the defect area. The evaluation of the results by radiography experts showed that this method has the efficiency to detect defects.