A Specific Task-Oriented Semantic Image Communication System for Substation Patrol Inspection
Senran Fan, Haotai Liang, Chen Dong, Xiaodong Xu, Geng Liu
Abstract
Intelligent inspection robots are widely used in substation patrol inspections to identify potential safety hazards by patrolling substations and sending back scene images. However, in areas with weak signals, the scene images may not be successfully transmitted, thereby reducing the quality of the robots' work. To reliably transmit high quality inspection images in weak signal areas, a Specific Task-oriented Semantic Communication System for Images (STSCI) is proposed. This system involves the extraction, transmission, restoration, and enhancement of semantic features to obtain clearer images sent by intelligent robots under weak signals. Inspired by the fact that only specific details of the image are required in substation patrol inspection tasks, we propose a new paradigm of semantic enhancement to ensure the clarity of key semantic information under low signal-to-noise ratio conditions. Experiments demonstrate that our STSCI generally outperforms traditional image-compression-based and channel-coding-based as well as other semantic communication systems in substation patrol inspection tasks under low signal-to-noise ratio conditions.