Automatic recognition of asbestos fibers

Asbestos is a generic name given to six fibrous minerals (chrysotile, crocidolite, amosite, anthophyllite, tremolite and actinolite asbestos), widely used in commercial products. Asbestos fibers are characterized by high tensile strength, flexibility, resistance to chemical attack and thermal degradation, large surface area and ability to be woven. These properties made the asbestos one of the most important material for cement, thermal insulation, sound proofing and anti-fire products. It has been produced also for use in textiles, gaskets, ropes, laminates with resins. The fibrous nature of asbestos minerals is dangerous: the light and thin fibers can spread in air and be respired, causing cancer, due to their chemical and physical properties. The use of asbestos fibers in coupling with complex materials make difficult its recognition. In airborne analysis, asbestos fibers can be confused with organic fibers such as cotton and wool. The aim of this work is to characterize asbestos fibers using image analysis as in previous work (Inoue et al., 1999), it is carried out in four steps: (i) image acquisition from airborne fibers samples; (ii) fibers identification by image segmentation; (iii) fibers analysis and features extraction, by dimension, shape, color and texture measurements; (iv) fibers classification.