A method based on color information and contour fragments was developed to identify citrus fruits in variable illumination conditions in the tree canopy, in order to guide the robots for harvesting citrus fruits. The color properties of target objects within natural citrus-grove scenes under various light conditions were analyzed, and a preliminary segmentation was put forward by fusing the Chromatic aberration information and normalized RGB model. The set of contour fragments was constructed via detecting the significant edge of Chromatic aberration map of R and B channels. The valid subset was selected by three parameters of the frament: length, bending degree and concavo-convex geometry characteristic. The ellipse fitting procedure was done to every frament, and the valid ones were chosen by the knowledge of fruit shape. The results showed that the occlusion contour were effectively recoveried under various light conditions using the proposed method, and the relative error of occlusion recovery was 5.34%.