MARKOV FUSION OF A PAIR OF NOISY IMAGES TO DETECT INTENSITY VALLEYS

Our presentation is related to a non-destructive control industrial task: the detection of defects on pairs of gamma-radiographic images. The images are very noisy and have a strong luminosity gradient. Defects are identified with intensity valleys. First we present a Bayes-Markov model in order to estimate the noise, the gradient and the valley bottom lines of defects for a single image. Then, we define a Markov fusion model for a pair incorporating a criterion of similarity between matched images. The proposed Markov models are general and can be used in other situations for detecting valley bottoms in noisy images.