Year-end Sale % OFF 
Abstract
Omnidirectional and 360° images are becoming widespread in industry and in consumer society, causing omnidirectional computer vision to gain attention. Their wide field of view allows the gathering of a great amount of information about the environment from only an image. However, the distortion of these images requires the development of specific algorithms for their treatment and interpretation. Moreover, a high number of images is essential for the correct training of computer vision algorithms based on learning. In this paper, we present a tool for generating datasets of omnidirectional images with semantic and depth information. These images are synthesized from a set of captures that are acquired in a realistic virtual environment for Unreal Engine 4 through an interface plugin. We gather a variety of well-known projection models such as equirectangular and cylindrical panoramas, different fish-eye lenses, catadioptric systems, and empiric models. Furthermore, we include in our tool photorealistic non-central-projection systems as non-central panoramas and non-central catadioptric systems. As far as we know, this is the first reported tool for generating photorealistic non-central images in the literature. Moreover, since the omnidirectional images are made virtually, we provide pixel-wise information about semantics and depth as well as perfect knowledge of the calibration parameters of the cameras. This allows the creation of ground-truth information with pixel precision for training learning algorithms and testing 3D vision approaches. To validate the proposed tool, different computer vision algorithms are tested as line extractions from dioptric and catadioptric central images, 3D Layout recovery and SLAM using equirectangular panoramas, and 3D reconstruction from non-central panoramas.
Highlights
The great amount of information that can be obtained from omnidirectional and 360o images makes them very useful
To evaluate if our synthetic images can be used in computer vision algorithms, we compare the evaluation of four algorithms with our synthetic images and real ones
We present a tool to create omnidirectional synthetic photorealistic images to be used in computer vision algorithms
Summary
The great amount of information that can be obtained from omnidirectional and 360o images makes them very useful. Being able to obtain information from an environment using only one shot makes these kinds of images a good asset for computer vision algorithms. To make faster and bigger datasets, previous works such as [1,2,3,4,5] use special equipment to obtain images, camera pose, and depth maps simultaneously from indoor scenes. These kinds of datasets are built from real environments, but need post-processing of the images to obtain semantic information or depth information. Tools like LabelMe [6] and new neural networks such as SegNet [7] can be used to obtain automatic semantic segmentation from the Sensors 2020, 20, 2066; doi:10.3390/s20072066 www.mdpi.com/journal/sensors
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
