There is a need to develop a non-destructive and fast detection method for bruising of fruits because the injuries lower quality of fruits, which lead to economic loss. In this paper, we propose a method to detect the bruise on apple surface with hyperspectral imaging technique. A hyperspectral image system consisting of a CCD digital camera, a line scanning spectrometer and a movable platform is designed to acquire the hyperspectral images of injured apples. Two models are established to distinguish the injured area on the surface from the normal area based on image processing technique with spatial clustering and Spectral Angle Mapper Classification (SAM), respectively. The discrimination accuracy of the SAM model is up to 100%, which is much higher than the spatial clustering model.