Statistical detection of defects in radiographic images using an adaptive parametric model

• Defect detection is addressed using testing theory and an original image model.
• An original model of non-anomalous background is proposed.
• The heteroscedastic property of noise due to photo-counting is considered.
• The statistical performance of the proposed test is analytical established.
• Numerical results show the accuracy of the proposed image model.

In this paper, a new methodology is presented for detecting anomalies from radiographic images. This methodology exploits a statistical model adapted to the content of radiographic images together with the hypothesis testing theory. The main contributions are the following. First, by using a generic model of radiographies based on the acquisition pipeline, the whole non-destructive testing process is entirely automated and does not require any prior information on the inspected object. Second, by casting the problem of defects detection within the framework of testing theory, the statistical properties of the proposed test are analytically established. This particularly permits the guaranteeing of a prescribed false-alarm probability and allows us to show that the proposed test has a bounded loss of power compared to the optimal test which knows the content of inspected object. Experimental results show the sharpness of the established results and the relevance of the proposed approach.