Detection and Extraction of Image Irregularities Using Morphological Pyramid of Detectors, Reconstruction, and Watershed

Mariusz Nieniewski
Electrotechnical Institute, and Institute of Fundamental Technological Research, Warsaw, Poland This work presents results obtained by the author using morphological methods for detection and extraction of image irregularities. The concept of irregularities typically refers here to small objects which have to be detected in the image. One area of experience has been detection and extraction of small defects of industrial products such as ferrite cores. The defects are detected using a morphological defect detectors arranged in the pyramid. This is a fast method which shows the locations of defects with rough estimate of their shape. The precise shape of small defects, visualized by their masks, is then obtained by a combination of morphological reconstructions. The presentation is illustrated by a large number of experimental examples of defect detection and extraction, as well as of extraction of microcalcifications.