A Z-shaped nonlinear transform for image segmentation and classification in intelligent debris analysis

In intelligent debris analysis, the large number of images to be processed requires significant time and memory resources. Moreover, image segmentation is difficult due to inconsistent illumination conditions. To address these problems, we propose a Z-shaped nonlinear transform for preprocessing that allows us to simultaneously obtain (a) a downsampled segmented image and (b) two novel features to be used for classification. Experimental results using a set of 93 wear particles show that our proposed method is efficient.