An Efficient Gyro-Aided Optical Flow Estimation in Fast Rotations With Auto-Calibration

Li, Ping; Ren, Hongliang

doi:10.1109/jsen.2018.2810060

Cited by 6 publications

(10 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The calibration data is shown Table 1 . Next, R ci was estimated by using the measured data from the gyroscope in the IMU sensor and the optical flow (see detail [ 16 ]). In addition, the average of R ci from 15 experiments with the CC+LS13 method in [ 16 ] are shown in Table 1 , which represents in term of quaternion.…”

Section: Resultsmentioning

confidence: 99%

“…The camera and the IMU sensor are calibrated as follows: (1) calibrate the camera to find out the focal length ( f ) by using the camera calibration module in OpenCV (Open Source Computer Vision); (2) calibrate the gyroscope to prevent gyro drift problems with averaging of the bias offset. We can assess the bias offset by measuring the output signal of the gyroscope over a long period of time when this sensor is sitting still and reduce noise by Kalman filter; (3) estimate R ci by using the relation between gyroscope data and optical flow that was proposed by Li and Ren [ 16 ] which is using the CC+LS13 method, and (4) determine the t off between the gyroscope and the camera input by a cross-correlation (CC) method [ 28 ] to synchronize operation between both of the sensors to accurately collect data in time. Then, we can correctly estimate the ω cam .…”

Section: Proposed Frameworkmentioning

confidence: 99%

“…The motion point on the next frame I imu ( p , q ) can be calculated from ω cam on the z -axis in (1) which is given by:

where φ equals to

divided by the frame rate of the input video, H and W are the height and width of the image, respectively. However, Li and Ren [ 16 ] performed the motion point with counteracting effect from the rotating motion in the z -axis, that is expressed as follows:

…”

Section: Motion Estimationmentioning

confidence: 99%

“…This tracker features success in evaluating the motion in the small movement but fails when exposed to a significant change in both local motion and global motion [ 15 ]. However, the large motion can be approximated by using IMU (Inertial Measurement Units) data as shown in [ 16 ] that demonstrates the efficiency of the gyroscope in estimating the optical flow during fast rotation. It adopted only the gyroscope data to aid the optical flow computation in improving the performance while measuring the motion estimations.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Hybrid Motion Estimation for Video Stabilization Based on an IMU Sensor

Auysakul

Pooneeth

2018

Sensors

View full text Add to dashboard Cite

Recorded video data must be clear for accuracy and faster analysis during post-processing, which often requires video stabilization systems to remove undesired motion. In this paper, we proposed a hybrid method to estimate the motion and to stabilize videos by the switching function. This method switched the estimated motion between a Kanade–Lucus–Tomasi (KLT) tracker and an IMU-aided motion estimator. It facilitated the best function to stabilize the video in real-time as those methods had numerous advantages in estimating the motion. To achieve this, we used a KLT tracker to correct the motion for low rotations and an IMU-aided motion estimator for high rotation, owing to the poor performance of the KLT tracker during larger movements. Furthermore, a Kalman filter was used to remove the undesired motion and hence smoothen the trajectory. To increase the frame rate, a multi-threaded approach was applied to execute the algorithm in the array. Irrespective of the situations exposed to the experimental results of the moving camera from five video sequences revealed that the proposed algorithm stabilized the video efficiently.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Proposed Frameworkmentioning

confidence: 99%

“…The motion point on the next frame I imu ( p , q ) can be calculated from ω cam on the z -axis in (1) which is given by:

where φ equals to

…”

Section: Motion Estimationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Hybrid Motion Estimation for Video Stabilization Based on an IMU Sensor

Auysakul

Pooneeth

2018

Sensors

View full text Add to dashboard Cite

show abstract

“…On the other hand, gyroscopes do not rely on image contents, which provide angular velocities in terms of roll, pitch, and yaw that can be converted into 3D motions, widely used for system control [26] and the HCI of mobiles [8]. Among all potential possibilities [2,28,14], one is to fuse the gyroscope for the motion estimation. Hwangbo et al proposed to fuse gyroscope to improve the robustness of KLT feature tracking [14].…”

Section: Introductionmentioning

confidence: 99%

GyroFlow: Gyroscope-Guided Unsupervised Optical Flow Learning

Li¹,

Luo²,

Liu³

2021

Preprint

View full text Add to dashboard Cite

Existing optical flow methods are erroneous in challenging scenes, such as fog, rain, and night because the basic optical flow assumptions such as brightness and gradient constancy are broken. To address this problem, we present an unsupervised learning approach that fuses gyroscope into optical flow learning. Specifically, we first convert gyroscope readings into motion fields named gyro field. Then, we design a self-guided fusion module to fuse the background motion extracted from the gyro field with the optical flow and guide the network to focus on motion details. To the best of our knowledge, this is the first deep learningbased framework that fuses gyroscope data and image content for optical flow learning. To validate our method, we propose a new dataset that covers regular and challenging scenes. Experiments show that our method outperforms the state-of-art methods in both regular and challenging scenes.

show abstract