Fig. 1. Diagram of improved adaptive depth image inpainting algorithm

Fig. 2. Image registration results acquired from Kinect v2 sensor. (a) Color image; (b) depth image

Fig. 3. Hole addition and inpainting results of partial depth images in Middlebury dataset. (a) Original color images; (b) original depth images; (c) depth images and details after adding holes to simulate noise; (d) depth images and details improved by proposed method

Fig. 4. Comparison of results obtained by proposed method and other methods. (a) Original color images; (b) original depth images; (c) results of joint bilateral filter; (d) results of Ref. [20]; (e) results of propsoed algorithm

Fig. 5. RMSE comparison of proposed algorithm and other methods based on the Middlebury dataset