The WHU Rolling Shutter Visual-Inertial Dataset

Abstract

The vast majority of modern consumer cameras employ a rolling shutter (RS) mechanism which has a price and electronic advantage to global shutter (GS). However, in geometric computer vision applications such as visual simultaneous localization and mapping (VSLAM), performances of accuracy and robustness are usually deteriorated due to the rolling shutter effect when using the RS cameras. This paper introduced the Wuhan University Rolling Shutter Visual-Inertial (WHU-RSVI) synthetic dataset for evaluating VSLAM and VI-SLAM (visual-inertial SLAM) methods in which RS cameras or IMU data are typically used. The proposed synthetic dataset contains RS images, time-synchronized GS images, inertial measurement unit (IMU) measurements, and accurate ground truth. It provides camera images with 640×480 resolution at 30 Hz and IMU measurements from 90 Hz to 14400 Hz. The cubic B-spline curves are used to model the motion of trajectories. Based on the known trajectories, an image of each pose can be rendered, and the corresponding IMU measurement model is then established. The dataset provides realistic images and IMU measurements by modeling the sensor noise in RGB and IMU data. Two trajectories with three sequences of different motion speeds (i. e., slow, medium and fast corresponding to different rolling shutter effects) are contained in the proposed dataset. Herein, the proposed dataset can be applied to compare the impact of different rolling shutter effects on a specific method.

Study on the Influence of Image Noise on Monocular Feature-Based Visual SLAM Based on FFDNet

“…However, the images in these datasets are all at low noise levels and have no different levels of noise. Therefore, a dataset for evaluating the noise of the visual SLAM method is extended from the previous publication WHU-RSVI [ 21 ]. However, the WHU-RSVI only provides a typical noise level, if the researchers want to study the noise of monocular visual SLAM, different noise levels and types are needed.…”

“…However, the WHU-RSVI only provides a typical noise level, if the researchers want to study the noise of monocular visual SLAM, different noise levels and types are needed. According to the noise model and clean images introduced in [ 21 ], 33 sequences with different noise levels and types are obtained, and the method to add image noise with customized noise types and levels is open-sourced. To reduce the impact of noise on the visual SLAM system, this paper uses the Fast and Flexible Denoising convolutional neural Network (FFDNet) [ 22 ] to denoise the image, and then the performance of original sequences and denoised sequences is evaluated.…”

“…The contributions of this paper are as follows: An open-source dataset extended from WHU-RSVI [ 21 ] for evaluating the noise of Monocular visual SLAM systems or denoising tasks. An open-source software for adding image noise with customized noise types and levels.…”

“…An open-source dataset extended from WHU-RSVI [ 21 ] for evaluating the noise of Monocular visual SLAM systems or denoising tasks.…”

“…To evaluate the influence of noise on the visual SLAM methods, a dataset with different levels of image noise is generated. The previous work of this dataset is WHU-RSVI [ 21 ], where the open-source ray-tracing software POV-Ray to generate images is used.…”