Adequate knowledge, such as information about the unique characteristics of each plant, is necessary to identify plant. Researchers have made plant recogni tion based on leaf characteristics. The leaf image- based plant recognition in view of different angles is a new challenge. In this study, the research on the p lant recognition was conducted based on leaf images resulted from 3D stereo camera. The 3D images are very influential in the development of computer vision t heory, which can provide more detailed information of an object. One of the information that can be obtai ned is about the position of the object in its imag e with the background as well as of the camera. One of the ways used to obtain such information is to calcula te the disparity. However, this method will only tell the position of the object compared to other objects wi thout that of range. Sum Absolute Different (SAD) is a method that can be used to find the disparity value. The SAD method does not require heavy computations and long process. Before calculating the disparity, all the images should be previously segmented. The objective of this segmentation is to separate all the objects from the background. Furthermore, filtering and polynomial transformation at the results of disparity is necess ary to improve the quality of resultant images. Furthermor e, 22 features were extracted using GLCM features (second order statistics) of images resulted from disparity improvement. The highest accuracy of match in the recognition of plant varieties was obtained at 50 cm distance a nd in the recognition of three plant varieties was 83.3%.