Accuracy improvement of counting asbestos in particles using a noise redacted background subtraction

Increased health damage caused by asbestos has become a problem recently. Removal of asbestos contained in building materials and rendering it harmless is a common means of alleviating asbestos hazards, but that process necessitates a judgment of whether asbestos is included in building materials. According to an official method, particles and asbestos must be counted in a sample to judge whether it contains asbestos. This work is performed visually and requires enormous amounts of time and effort. Consequently, automated counting using background subtraction is proposed for rapid, highly accurate analysis. However, the method does not enable accurate counting because of noise included in a background image. This study is intended to improve the accuracy of counting particles through noise removal using a Gaussian filter.