Stereo Matching for Binocular Citrus Images Using Surf Operator and Epipolar Constraint

We presented an application of SURF operator and epipolar constraint for stereo matching of binocular citrus images. First, R-B space images were achieved by linear transform of original images´ color space. Second, fast Hessian-matrix detector was used to detect the interest points and the SURF descriptor was used to describe the points´ features. Finally, Euclidean distance and the epipolar constraint were used for features matching. Experiment shows that the average execution time of the over-all algorithms (including the feature extraction and the feature matching) is 293ms.The method can detect and match interest points accurately without the influence of fruits´ occlusions. Once stereo matching work has done successfully, the following computation of depth information of the fruits may be applied.