Image processing methods based on slope equalization and two-dimensional operators

Digital x-ray radiography (DR) is replacing classical x-ray film radiography, and is becoming one of the main volumetric inspection and diagnostics methods. DR has many advantages over classical film radiography, one of which is the possibility to apply image processing methods. This can lead to improved analysis and interpretation of the radiographic images, and offers the possibility for automation of the inspection process, and for use as a image pre-processing before applying machine learning and artificial intelligence. This work describes the development and evaluation of a new image processing method, based on slope equalization. This is a non-linear method, which uses equalized images of the slopes in the original image, and thus tries to equalize the visibility of details with low and high contrast. In addition, two-dimensional (2D) operators for calculating smoothed and derivative values (slopes) for each pixel were developed and are also presented and evaluated. The use of smoothed 2D derivative operators for enhancing visibility of details in radiographic images was demonstrated.