Many medical images contain objects with characteristic shape and texture. Detection and classification of regional outlines can be made more robust if specific texture descriptors are established and then matched in an appropriate feature space. The authors develop a design approach which, after training, is able to discriminate between regions with structured texture. They then apply the same general method to images with random texture. (6 pages)
Inspec keywords: medical image processing; filtering theory; genetic algorithms; image texture; image recognition
Subjects: Optimisation techniques; Biology and medical computing; Medical and biomedical uses of fields, radiations, and radioactivity; health physics; Optimisation techniques; Optical information, image and video signal processing; Radiation and radioactivity applications in biomedicine; Patient diagnostic methods and instrumentation; Computer vision and image processing techniques
Related content
content/conferences/10.1049/ic_19970127
pub_keyword,iet_inspecKeyword,pub_concept
6
6
