Laser & Optoelectronics Progress, Volume. 60, Issue 24, 2410012(2023)
Curved Texture Flattening Algorithm Based on the Light Field Camera
Fig. 1. Flow chart of the curved texture flattening algorithm
Fig. 2. Schematic of local texture distortion correction
Fig. 3. Simulated output of light field camera for the curved characters texture. (a) Texture image; (b) depth map
Fig. 4. Surface segmentation. (a) Cluster evaluation value; (b) cluster result; (c) local texture images before and after dilation
Fig. 5. Correction results of local textures. (a)‒(d) Local textures before correction; (e)‒(h) local textures after correction
Fig. 6. Image stitching. (a) Stitching of the local texture images in Fig. 5; (b) texture flattening result of the whole curved surface
Fig. 7. Flattening results of different texture images. (a)‒(c) Original texture images of fingerprint, chinaware and mural; (d)‒(f) texture images of fingerprint, chinaware and mural on curved surface; (g)‒(i) flattening results of curved fingerprint, chinaware and mural
Fig. 8. Texture flattening experiment based on the focused light field camera. (a) Curved finger image output by light field camera; (b) depth map output by light field camera; (c) flattening result of the curved finger image
Fig. 9. Flattening results for textures on surface with different degrees of bending (the image at the lower-right corner is a close-up of the region labeled by rectangular box). (a)‒(c) Texture images on curved surface with different degrees of bending; (d)‒(f) corresponding texture flattening results
Fig. 10. Flattening results of noisy texture images with different contrasts (the image at lower-right corner is a close-up of the region labeled by the rectangular box). (a)‒(d) Curved texture images with 0.2‒0.5 contrasts; (e)‒(h) corresponding texture flattening results (contrasts of 0.2‒0.5); (i)‒(l) curved texture images with 0.6‒0.9 contrasts; (m)‒(p) corresponding texture flattening results (contrasts of 0.6‒0.9)
Fig. 11. Perspective correction results of the local texture images with depth noise of different standard deviations. (a) Texture image (the region labeled with rectangular box is the region to be corrected); (b)
Fig. 12. Texture image and its flattening result when its depth map is superimposed with 0.005 mm Gaussian noises. (a) Texture image (the region labeled with rectangular box is the local texture selection range); (b) flattening result
Fig. 13. Correction results of local texture images with different normal vectors. (a)~(e) Local textures to be corrected; (f)~(j) correction results of local texture images without depth noise; (k)~(o) correction results of local texture images with depth noise of the same standard deviation
|
|
Get Citation
Copy Citation Text
Shengnan Qin, Yanting Lu. Curved Texture Flattening Algorithm Based on the Light Field Camera[J]. Laser & Optoelectronics Progress, 2023, 60(24): 2410012
Category: Image Processing
Received: Mar. 23, 2023
Accepted: May. 15, 2023
Published Online: Nov. 27, 2023
The Author Email: Lu Yanting (ytlu@niao.ac.cn)