The semi-global matching algorithm based on slanted plane smoothing can achieve good results in disparity discontinuous area.It can reduce disparity fracture phenomenon and makes the disparity image more smoothing.However

the slanted plane smoothing algorithm has huge computing quantity and long run time.If the input images are not very ideal

the final disparity image could have black small blocks.Aim at the above problems

binocular stereo vision algorithm base on improved slanted plane smoothing is proposed in this paper.The proposed algorithm not only increases the speed of calculation

but also fills the invalid disparity in original disparity image using the nearest interpolation algorithms.Considering that the gradients of the pixels in the same segmentation should be close

the gradient energy is supplemented to optimize the energy function.So the proposed algorithm segments the image more reasonably.The experiments demonstrate that the proposed algorithm has a better disparity result.The subjective quality is improved

and the time consuming is also increased by 40%.