Crane is important lifting equipment for material in the production. Studying the safe, reliable and efficient operation of the crane is of great significance. In order to study the operation characteristics of crane, and the feasibility of avoiding obstacles and monitoring operation based on visual system, a crane robot and its visual measurement system in laboratory environment were built.The motion trajectory coordinate system of the crane hook and binocular stereo vision model were established.Different color threshold selection method for RGB was used to identify the hook target image, and then hook was separated from the complex background.Finally, the SIFT feature point matching method was used to realize the matching of the feature points and the visual detection and tracking for the moving hook. The research results may offer a reference to improve the automation and intelligence of crane operation.