Machine vision inspection of white foreign fibers in cotton is a research hotspot in China and aboard. Based on the fact that there are flosses and free filaments on the surface of a cotton layer，and most of the white foreign fiber surfaces are smooth, without any floss， an image acquisition method of the cross-section of the samples under the illumination of linear laser was proposed, which showed the differences of the microstructure characteristic of their surface from the reflection and transmission of the linear laser. Then, according to the features of the distribution and density of the flosses around the white spots in the images, an algorithm for identification of cotton and white foreign fibers was suggested. The experimental results indicated that, by the new method, the detecting rate of the white foreign fibers in cotton was up to 86.9%.