Project Description

Considering the great demand for reducing human exhaustion and to enhance POD in the domain of NDT, several methods are being evolved. However, the accuracy and time to complete the detection and interpretation of defects depends on the type of methods that are being followed. To improve the process, an attempt towards incorporating several computational techniques is being evolved. Yet, a robust model incorporating efficient image processing techniques and a strong interpretation mechanism with an appropriate visualization technique is commanded. Hence, a self-directed dashboard for processing industrial radiographs is proposed. The system attempts to process the radiographs to segregate and extract the different ROIs (defects) from the given image using an automated task. Subsequently, the image features such as geometrical features, statistical features and textural features are extracted which is in combination, interpreted and classified using a machine learning model. Thus a complete AI based computational solution is devised for processing, interpreting the electron beam weld defects of Spacecraft and launch vehicle components.