Adaptive ORB feature detection with a variable extraction radius in RoI for complex illumination scenes

Wu, Xing; Sun, Chao; Chen, Leisheng; Zou, Ting; Yang, Wei; Xiao, Haining

doi:10.1016/j.robot.2022.104248

Cited by 6 publications

(1 citation statement)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[20] Similar to K-Means' approximate clustering algorithm, PROSAC is commonly used for edge detection, which improves the speed and accuracy of matching by continuously repeating the sampling and clustering process. [21] When used for image matching error rejection, the efficiency is more than double that of RANSAC, and it has good robustness. Ma Xiaomin et al [22] proposed a method to construct a nonlinear scale space with complete affine invariance by combining asymptotic sampling optimization and nonlinear diffusion filtering, and improved the matching speed and accuracy through vector field consistency (VFC) and progressive sampling consistency (PROSAC).…”

Section: Asymptotic Consistent Samplingmentioning

confidence: 99%

Dynamic SLAM algorithm adopt with eliminating mismatched point chains in grid motion statistics

He,

Yu,

2024

Preprint

View full text Add to dashboard Cite

Feature matching is an essential part in areas such as target tracking, and three-dimensional reconstruction. In case of rotational motion in the image, the rotating exercise core 8 statistical motion support volume is applied, resulting in low matching accuracy and long time to eliminate mismatching. A principal component analysis method is proposed to calculate rotation angle, feature points are changed in the grid and its neighborhood grid, which sets Gaussian threshold according to Euclid distance between neighborhood feature point and the matching feature point. And a new fractional statistical model is proposed to increase the number of correct matching pairs, So as to improve the fastness and accuracy of characteristic matching. Aiming at the problem of mismatch caused by local similarity of images, a data set is proposed to determine the data set by using geometric relationship between feature points, which analyzes the similarity between the data by the Person correlation coefficient, and sets the threshold to remove the feature matching pairs with low confidence, so as to improve the accuracy of feature matching. Experimental results show that the feature matching speed of the improved GMS algorithm is 3 times that of original GMS algorithm, and the false matching is eliminated in local similar region, which improves the quality of feature matching.

show abstract

Section: Asymptotic Consistent Samplingmentioning

confidence: 99%

Dynamic SLAM algorithm adopt with eliminating mismatched point chains in grid motion statistics

He,

Yu,

2024

Preprint

View full text Add to dashboard Cite

show abstract

Automated dimensional quality inspection of super-large steel mesh using fixed-spacing detection transformer and improved oriented fast and rotated brief

Guo,

Huang,

Zhang

et al. 2025

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

An intelligent vision recognition method based on deep learning for pointer meters

Chen

Sun

et al. 2023

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Nowadays, pointer instruments remain the main state monitoring devices in the power industry, because they have strong mechanical stability to resist electromagnetic interferences compared with digital instruments. Although the object detection algorithms based on deep learning have widely been used in the field of instrument detection, the meter recognition process still relies on threshold segmentation to recognize object points and on Hough transform to extract the meter pointer. An intelligent vision recognition method based on YOLOv5 and U2-Net network (YLU2-Net) is proposed to improve the accuracy and efficiency of meter recognition in a complex environment. Firstly, the pointer meter is located in the instrument images by using the YOLOv5 network as a Region of Interest (RoI). Then, the instrument RoI is processed by means of perspective transformation and image resizing. Thirdly, an improved U2-Net image segmentation method with the deep separable convolution and the focal loss function is devised to distinguish the pointers and scales from the background in the instrument RoI. Further, a dimension reduction reading method with the polar coordinate transformation is developed to calculate the meter reading accurately and efficiently. Finally, the ablation experiment is conducted to test the performance of each algorithm module in our method, and the competition experiment is completed to compare our method with other state-of-the-art ones. The experimental results verify the accuracy and efficiency of the YLU2-Net recognition method proposed.

show abstract

Adaptive ORB feature detection with a variable extraction radius in RoI for complex illumination scenes

Cited by 6 publications

References 32 publications

Dynamic SLAM algorithm adopt with eliminating mismatched point chains in grid motion statistics

Dynamic SLAM algorithm adopt with eliminating mismatched point chains in grid motion statistics

Automated dimensional quality inspection of super-large steel mesh using fixed-spacing detection transformer and improved oriented fast and rotated brief

An intelligent vision recognition method based on deep learning for pointer meters

Contact Info

Product

Resources

About