Small target recognition method on weak features
- PDF / 1,474,012 Bytes
- 19 Pages / 439.37 x 666.142 pts Page_size
- 69 Downloads / 232 Views
Small target recognition method on weak features QingE Wu 1
1
1
1
& Ziming An & Hu Chen & Xiaoliang Qian & Lijun Sun
2,3
Received: 6 April 2020 / Revised: 13 September 2020 / Accepted: 17 September 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
For real time tracking the moving small target under complex backgrounds, this paper will present a tracking algorithm and a recognition method for small target based on the fusion filtering of single frame and multiple frames. For a case that the moving small target is masked or submersed easily by other objects or noise in complex background, this paper will present an opening and closing transform algorithm to eliminate or weaken background and noise; present a competitive model with adaptive neural network of online learning for the puniness characteristics of moving small target, use its competitive active unit to extract the multidimensional characteristic parameter of small target; consequently, present a tracking recognition method for moving small target based on a complex background. In conclusion, the significant novelty is the small target recognition method proposed, including the eliminating noise method, extraction method for weak small features, establishment of prediction model of motion target state, recognition method for small target. These researches development in this paper help to the personnel of target recognition and image processing understand the motion law, reduce expenditure, decrease unnecessary damage or injury, etc. Keywords Small target detection . Small target recognition . Track tracking . Multi-frame chain . Fusion
* QingE Wu [email protected]
1
School of Electrical and Information Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China
2
Key Laboratory of Grain Information Processing and Control, Ministry of Education, Henan University of Technology, Zhengzhou 450001, China
3
Zhengzhou Key Laboratory of Machine Perception and Intelligent System, Henan University of Technology, Zhengzhou 450001, China
Multimedia Tools and Applications
1 Introduction The detection and extraction of small target signals with low signal-to-noise ratio is one of the key technical problems to be solved by various advanced detection systems, the target image occupies only one or several pixels of the area, and because the background environment is complex, the unevenness of atmospheric radiation, the detector’s internal noise and so on factors, the target is almost covered in the clutter background, no Shape and structure information, and sometimes may even lose the target, which gives a small target detection has brought great difficulty. The detection problem of low signal-to-noise ratio under strong clutter background conditions directly determines the operating distance and detection performance of the detection system, and its solution has very important practical significance for improving the performance of the detection system. In order to effectively detect the low signal-to-no
Data Loading...
