Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images

Nie, Lang; Lin, Chunyu; Liao, Kang; Liu, Shuaicheng; Zhao, Yao

doi:10.1109/tip.2021.3092828

Cited by 158 publications

(115 citation statements)

References 40 publications

Supporting

Mentioning

115

Contrasting

Order By: Relevance

“…Global Homograp ĥy [16] Graphcut APAP [21] PTIS [22] ANAP [23] NIS [24] REW [25] Perazzi [27] UDIS [34] Alignment over an/along a . .…”

Section: Methodsmentioning

confidence: 99%

“…This is later referred to as a “non-parallax” view. A very recent article proposed an unsupervised deep image stitching framework [ 34 ], later referred to as the UDIS method, for image stitching that does not require a supervised synthetic dataset. The UDIS framework is composed of: (a) an initial neural network for the global homography estimation and (b) a second architecture for alignment refinement with reconstruction networks.…”

Section: State-of-the-art In Image Stitchingmentioning

confidence: 99%

“…Due to the lack of a standard quantitative metric, as explained in Section 2 , the evaluation of the different algorithms was performed in relation to these problems. We implemented a module on the CamiTK platform [ 35 ] that could test multiple image stitching algorithms: Global homography [ 16 ]; Graphcut (global homography and graph cut [ 19 ]); Mesh-based models: APAP [ 21 ], ANAP [ 23 ], NIS [ 24 ], REW [ 25 ] and PTIS [ 22 ]; Two non-mesh-based models: the optical flow-based model Perazzi [ 27 ] and the neural network-based model UDIS [ 34 ]. …”

Section: Benchmarking Of Image Stitching Algorithmsmentioning

confidence: 99%

“…Two non-mesh-based models: the optical flow-based model Perazzi [ 27 ] and the neural network-based model UDIS [ 34 ].…”

Section: Benchmarking Of Image Stitching Algorithmsmentioning

confidence: 99%

“…Global homography [16]; • Graphcut (global homography and graph cut [19]); • Mesh-based models: APAP [21], ANAP [23], NIS [24], REW [25] and PTIS [22]; • Two non-mesh-based models: the optical flow-based model Perazzi [27] and the neural network-based model UDIS [34].…”

mentioning

confidence: 99%

See 4 more Smart Citations

Qualitative Comparison of Image Stitching Algorithms for Multi-Camera Systems in Laparoscopy

et al. 2022

View full text Add to dashboard Cite

Multi-camera systems were recently introduced into laparoscopy to increase the narrow field of view of the surgeon. The video streams are stitched together to create a panorama that is easier for the surgeon to comprehend. Multi-camera prototypes for laparoscopy use quite basic algorithms and have only been evaluated on simple laparoscopic scenarios. The more recent state-of-the-art algorithms, mainly designed for the smartphone industry, have not yet been evaluated in laparoscopic conditions. We developed a simulated environment to generate a dataset of multi-view images displaying a wide range of laparoscopic situations, which is adaptable to any multi-camera system. We evaluated classical and state-of-the-art image stitching techniques used in non-medical applications on this dataset, including one unsupervised deep learning approach. We show that classical techniques that use global homography fail to provide a clinically satisfactory rendering and that even the most recent techniques, despite providing high quality panorama images in non-medical situations, may suffer from poor alignment or severe distortions in simulated laparoscopic scenarios. We highlight the main advantages and flaws of each algorithm within a laparoscopic context, identify the main remaining challenges that are specific to laparoscopy, and propose methods to improve these approaches. We provide public access to the simulated environment and dataset.

show abstract