Research Article
Edge tracking method of damaged mural images based on deep learning
@INPROCEEDINGS{10.4108/eai.27-8-2020.2295946, author={Li-wei ZHANG and Huan-ping FENG}, title={Edge tracking method of damaged mural images based on deep learning}, proceedings={Proceedings of the 13th EAI International Conference on Mobile Multimedia Communications, Mobimedia 2020, 27-28 August 2020, Cyberspace}, publisher={EAI}, proceedings_a={MOBIMEDIA}, year={2020}, month={11}, keywords={deep learning; image edge; edge tracking;}, doi={10.4108/eai.27-8-2020.2295946} }- Li-wei ZHANG
Huan-ping FENG
Year: 2020
Edge tracking method of damaged mural images based on deep learning
MOBIMEDIA
EAI
DOI: 10.4108/eai.27-8-2020.2295946
Abstract
The traditional edge tracking method of damaged mural images has the phenomenon of high noise variance in the process of tracking, which affects the actual performance of tracking. Therefore, an edge tracking method based on deep learning is proposed.Through image preprocessing, the image gray mean value was unified and image quality was enhanced. Roberts edge detection operator was used to detect image edge features. RCF model in deep learning was used to fuse edge features and output them to find the starting point of fusion features.The test results show that the noise variance of the edge tracking method for damaged mural images designed based on deep learning is between 0.010 and 0.015, which is lower than the noise variance generated in the traditional tracking method, indicating that this method is better than the traditional tracking method.
