Abstract:Compared to contact sensors-based motion measurement, vision-based motion measurement has advantages of low cost and high efficiency and have been under active development in the past decades. This paper provides a review on existing motion measurement methods. In addition to the development of each branch of vision-based motion measurement methods, this paper also discussed the advantages and disadvantages of existing methods. Based on this discussion, it was identified that existing methods have a common limitation in optimally balancing accuracy and robustness. To address issue, we developed the Gaussian kernel-based motion measurement method. Preliminary study shows that the developed method can achieve high accuracy on simple synthesized images.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is that it is difficult for the existing vision - based motion measurement methods to achieve the optimal balance between accuracy and robustness. Specifically: 1. **Match - based methods**: These methods have high robustness, but perform poorly in sub - pixel - level accuracy. Since template matching is carried out at the pixel level, it is difficult to achieve high - precision motion estimation. 2. **Gradient - based methods**: These methods can achieve high precision, but are very sensitive to parameter selection, resulting in low robustness. Different texture patterns will affect the accuracy of these methods. To solve the above problems, the author has developed a Gaussian kernel - based motion measurement method. This method represents an image by using a set of Gaussian kernels and measures the motion of the target object by using the position changes of these kernels. Preliminary research shows that this method can achieve high - precision motion estimation on simple synthetic images. ### Specific problem description - **Limitations of match - based methods**: - High robustness but low accuracy, especially performing poorly in sub - pixel - level motion estimation. - Only able to match at the pixel level, it is difficult to achieve sub - pixel - level accuracy. - **Limitations of gradient - based methods**: - High precision but sensitive to parameter selection, resulting in low robustness. - Different texture patterns will affect the accuracy of the method and it is difficult to adapt to all situations. ### Solution The author proposes a Gaussian kernel - based motion measurement method, aiming to improve both accuracy and robustness simultaneously. The core ideas of this method are: - **Gaussian kernel representation**: Represent each image as a set of Gaussian kernels, and an image sequence is represented by the position changes of these kernels. - **Optimized balance**: By optimizing the position and parameters of the Gaussian kernels, this method can maintain high precision and robustness under different texture patterns. ### Preliminary research results Preliminary research shows that this method can achieve a motion estimation of 0.01 pixels on simple synthetic images, with an average error of \(3\times10^{- 4}\) pixels. This indicates that the Gaussian kernel - based motion measurement method has good potential, and its performance can be improved in the future by further optimizing initialization, the optimization process, parameter selection techniques and loss functions, etc. ### Summary This paper reviews the existing vision - based motion measurement methods, points out the balance problem between accuracy and robustness in the existing methods, and proposes a Gaussian kernel - based motion measurement method to solve this problem. Preliminary research shows that this method performs well on simple synthetic images and is expected to achieve better results in practical applications in the future.

A review on vision-based motion estimation

Moving Object Tracking Research Based on Active Vision

Motion estimation based feature selection for visual SLAM

A New Method for Increasing Accuracy of Distance Measurement Based on Single Visual Camera

Realization of odometry system using monocular vision

Robust and Efficient Vehicles Motion Estimation with Low-Cost Multi-Camera and Odometer-Gyroscope

A Survey on the Vision-Based Human Motion Recognition

Visual Odometry in dynamical scenes

A Review of Visual Odometry Methods and Its Applications for Autonomous Driving

Vision-based automatic structural motion estimation in presence of 3D camera motion effects

Vision-based Structural Displacement Measurement: System Performance Evaluation and Influence Factor Analysis

Feature Regions Segmentation Based RGB-D Visual Odometry in Dynamic Environment

Vision-based Displacement Measurement Sensor Using Modified Taylor Approximation Approach

Time-Varying Motion Filtering for Vision-Based Nonstationary Vibration Measurement

A Study of Vision based Human Motion Recognition and Analysis

A High-Speed Vision-Based Sensor for Dynamic Vibration Analysis Using Fast Motion Extraction Algorithms.

A Survey of Visual Analysis of Human Motion and Its Applications

Efficient visual odometry estimation using stereo camera

A new vision measurement method based on active object gazing

Robot motion visual measurement based on RANSAC and weighted constraints method

Motion Capture from Inertial and Vision Sensors