Weed management is vitally important in crop production systems. However, conventional herbicide-based weed control can lead to negative environmental impacts. Manual weed control is laborious and impractical for large scale production. Robotic weeding offers a possibility of controlling weeds precisely, particularly for weeds growing close to or within crop rows. The fusion of two-dimensional textural images and three-dimensional spatial images to recognize and localize crop plants at different growth stages were investigated. Images of different crop plants at different growth stages with weeds were acquired. Feature extraction algorithms were developed, and different features were extracted and used to train plant and background classifiers, which also addressed the problems of canopy occlusion and leaf damage. Then, the efficacy and accuracy of the proposed methods in classification were demonstrated by experiments. Currently, the algorithms were only developed and tested for broccoli and lettuce. For broccoli plants, the crop plants detection true positive rate was 93.1%, and the false discover rate was 1.1%, with the average crop-plant-localization error of 15.9 mm. For lettuce plants, the crop plants detection true positive rate was 92.3%, and the false discover rate was 4.0%, with the average crop-plant-localization error of 8.5 mm. The results have shown that 3D imaging based plant recognition algorithms are effective and reliable for crop/weed differentiation.