Fig. 1. Depth images obtained from 3D light field capture system

Fig. 2. Process of image stitching

Fig. 3. Bilateral filter of depth image and comparison results before and after interpolation. (a) Depth original image; (b) after bilateral filter; (c) after interpolation

Fig. 4. Corners detection results based on depth image.(a) Harris corner; (b) proposed algorithm

Fig. 5. Features point matching measurement algorithm

Fig. 6. Stitching results on different dataset. (a) Open dataset on the web; (b) dataset captured by camera; (c) dataset captured in micro-camera

Fig. 7. Matching results in depth image without noise. Results of the proposed algorithm after (a1) first feature matching and (a2) secondary feature matching screening; results of SIFT algorithm after (b1) first feature matching and (b2) secondary feature matching screening

Fig. 8. Matching results in depth image with noise. Results of the proposed algorithm after (a1) first feature matching and (a2) secondary feature matching screening; results of SIFT algorithm after (b1) first feature matching and (b2) secondary feature matching screening

Fig. 9. Stitching results of depth grayscale image based on different algorithms. (a) Homography algorithm; (b) SIFT and kd-tree algorithm; (c) robust elastic warping algorithm; (d) cylindrical warping and SIFT algorithm; (e) ORB algorithm; (f) proposed algorithm