Two-dimensional transition metal sulfides (TMDCs) are widely used in solar cells, photocatalysis, sensors, flexible electronic devices and other fields due to their unique exciton effect and material properties. The layer-number has a significant effect on their properties, thus rapid and automatic detection technology of TMDC samples with required layer-number is urgently required for their application expanding from laboratories to semiconductor manufacturing industries. In this paper, a thin-layer TMDC sample automatic detection technology was proposed, combining the hyperspectral imaging method and the image processing algorithm. The optical contrast of TMDC samples with different layer-numbers was systematically studied using a self-built reflection hyperspectral imaging system. The differential reflection mechanism of the optical contrast was elucidated, and a reliable layer-number determination method was proposed. Additionally, an image processing algorithm was designed to seek out TMDC samples with determined layer-number from the microscopic images. This method is universal and practical which can realize large-scale automatic sample detection when combining with auto-focus scanning control. It also provides new inspiration for detection of other surface targets at microscopic view.