Near-duplicate image removal method for aerial transmission line inspection

UAV inspection has become an important way of transmission line inspection. However, the large number of inspection images collected by UAVs presents a series of challenges for subsequent processing. Aiming at the problem that there are a large number of approximate duplicate images in the inspectio...

Full description

Saved in:
Bibliographic Details
Main Authors: Li, Huixin, Guo, Yangming, Liu, Yuhang, Peng, Junsheng, Yang, Chao
Format: Conference Proceeding
Language:English
Subjects:
Autonomous aerial vehicles
Clustering algorithms
Feature extraction
hash index
Indexes
Industrial electronics
Inspection
near-duplicate image
Power transmission lines
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:UAV inspection has become an important way of transmission line inspection. However, the large number of inspection images collected by UAVs presents a series of challenges for subsequent processing. Aiming at the problem that there are a large number of approximate duplicate images in the inspection data collected by UAV. This paper proposes a hierarchical clustering-based near-duplicate image deletion method that combines a "coarse grouping" strategy based on block feature extraction and a "fine grouping" strategy based on the Sobel operator of perceptual hashing. The hash index structure is constructed separately in two stages to improve matching efficiency. In the near-duplicate image removal stage, the content and attributes of the images are comprehensively evaluated to accurately identify the centroid image. Experimental results demonstrate that this method effectively removes near-duplicate images while improving overall efficiency and ensuring image deduplication accuracy. Thus, it is an effective method for near-duplicate image removal.
ISSN:2158-2297
DOI:10.1109/ICIEA58696.2023.10241777