Aiming at the disadvantage of the small monitoring range of single camera,an image mosaic method based on binocular cameras is presented. Firstly,the interesting points from video sequence image are detected,and improved scale invariant feature transform algorithm is introduced to build the 64-dimensional descriptors. In the process of matching,BBF algorithm is adopted,which can improve the matching efficiency. After that,the error matching is eliminated by using random sample consensus method. Finally,the image mosaic is done with the weighted average method. The experimental results show that this method has a higher matching efficiency and accuracy,and it can effectively extend monitoring range. It has a practical application value in the fields of video surveillance,information collection and judicial forensics.