End-to-End Blurry Template Matching Method Based on Siamese Networks

A Robust and Accurate End-to-End Template Matching Method Based on the Siamese Network

IEEE Geoscience and Remote Sensing Letters ◽

10.1109/lgrs.2021.3094046 ◽

2021 ◽

pp. 1-5

Author(s):

Qiang Ren ◽

Yongbin Zheng ◽

Peng Sun ◽

Wanying Xu ◽

Di Zhu ◽

...

Keyword(s):

Template Matching ◽

Matching Method ◽

Siamese Network ◽

End To End

Download Full-text

Video-Based Person Re-Identification by an End-To-End Learning Architecture with Hybrid Deep Appearance-Temporal Feature

Sensors ◽

10.3390/s18113669 ◽

2018 ◽

Vol 18 (11) ◽

pp. 3669 ◽

Cited By ~ 3

Author(s):

Rui Sun ◽

Qiheng Huang ◽

Miaomiao Xia ◽

Jun Zhang

Keyword(s):

Feature Representation ◽

Visual Sensor ◽

Stream Structure ◽

Temporal Features ◽

Feature Structure ◽

End To End ◽

Global And Local ◽

Siamese Networks ◽

Appearance Feature ◽

Temporal Feature

Video-based person re-identification is an important task with the challenges of lighting variation, low-resolution images, background clutter, occlusion, and human appearance similarity in the multi-camera visual sensor networks. In this paper, we propose a video-based person re-identification method called the end-to-end learning architecture with hybrid deep appearance-temporal feature. It can learn the appearance features of pivotal frames, the temporal features, and the independent distance metric of different features. This architecture consists of two-stream deep feature structure and two Siamese networks. For the first-stream structure, we propose the Two-branch Appearance Feature (TAF) sub-structure to obtain the appearance information of persons, and used one of the two Siamese networks to learn the similarity of appearance features of a pairwise person. To utilize the temporal information, we designed the second-stream structure that consisting of the Optical flow Temporal Feature (OTF) sub-structure and another Siamese network, to learn the person’s temporal features and the distances of pairwise features. In addition, we select the pivotal frames of video as inputs to the Inception-V3 network on the Two-branch Appearance Feature sub-structure, and employ the salience-learning fusion layer to fuse the learned global and local appearance features. Extensive experimental results on the PRID2011, iLIDS-VID, and Motion Analysis and Re-identification Set (MARS) datasets showed that the respective proposed architectures reached 79%, 59% and 72% at Rank-1 and had advantages over state-of-the-art algorithms. Meanwhile, it also improved the feature representation ability of persons.

Download Full-text

A multiple-layer flexible mesh template matching method for non-rigid registration between a pelvis model and CT images

10.1117/12.481354 ◽

2003 ◽

Cited By ~ 4

Author(s):

Jianhua Yao ◽

Russell H. Taylor

Keyword(s):

Template Matching ◽

Ct Images ◽

Matching Method ◽

Rigid Registration ◽

Multiple Layer

Download Full-text

Dynamic target template matching for railway catenary suspension motion detection in wind area

International Journal of Distributed Sensor Networks ◽

10.1177/1550147718797956 ◽

2018 ◽

Vol 14 (9) ◽

pp. 155014771879795 ◽

Cited By ~ 1

Author(s):

Wei Zhou ◽

Heting Xiao ◽

Zhonggang Wang ◽

Lin Chen ◽

Shaoqing Fu

Keyword(s):

Template Matching ◽

Three Dimensional ◽

Moving Target ◽

Maximum Relative Error ◽

Gradual Change ◽

Matching Method ◽

Safety Research ◽

Target Template ◽

Target Templates ◽

Dynamic Target

A dynamic target template matching method was proposed to identify railway catenary suspension movements of wind-induced vibration in wind area. Catenary positioning point was taken as the target template, which was compared with equal-sized image sequentially using the proposed matching difference. And, three-dimensional contour map of matching difference value at each sub-area was obtained, where the target pixel coordinates were determined by the minimum matching difference value. Considering the complex imaging condition, the target template was updated by the detected target image to sense the gradual change of illumination conditions like brightness and contrast. Furthermore, to eliminate detecting errors due to wind-induced camera vibration, both static and moving target templates were identified for acquiring the absolute motion of the moving target. Finally, validation test was performed with animation in PowerPoint. The calculated target displacement agrees well with theoretical motion with maximum relative error of 1.8%. And experiment application was conducted at site by analyzing the relationship between detecting displacement and wind speed. Results indicate that the proposed dynamic target template matching method can meet required engineering precision and provide an effective way for wind-vibration safety research of railway catenary system in wind area.

Download Full-text