A post processing algorithm for upsampling depth image based on boundary correction

Depth image is widely used in the fields such as human-computer interaction, navigation, and augmented reality. However, the depth images delivered by current depth cameras such as Kinect or Time of Flight cameras are limited in lower resolution compared to their corresponding color images, thus they cannot meet the requirements of practical applications that require depth images to have the same resolution of color images. To address this problem, this paper proposed a post processing algorithm for upsampling depth image based on boundary correction. Under this algorithm, after preliminarily upsampling of low resolution depth images by classic interpolation algorithm, two post-processing steps were carried out on the preliminarily up-sampled depth images. In the next step, the geometrical edge from the color image was detected so as to guide the boundary correction of up-sampled depth image. Then, the pixel values in central smooth region of depth image were extracted to fill the corrected depth edge region of the up-sampled depth images. The experimental results showed that the proposed algorithm generated higher quality upsampling depth image, compared with classic interpolation algorithm, and eliminated fuzzy boundaries brought about by classic interpolation algorithm under the condition of smooth boundary area. The edge of depth image matched better the geometric edge of color image through proposed boundary correction, resulting in the improvement of the quality of the synthetic viewpoint. © 2015, Science Press. All right reserved.