A Deep Single Image Rectification Approach for Pan-Tilt-Zoom Cameras
This work addresses image quality issues for PTZ camera users in surveillance and visual applications, representing an incremental improvement over existing deep learning methods.
The paper tackles the problem of rectifying nonlinear distortions in wide-angle Pan-Tilt-Zoom (PTZ) camera images, which is crucial for surveillance, by introducing FDBW-Net, a deep learning framework that achieves state-of-the-art performance in distortion rectification.
Pan-Tilt-Zoom (PTZ) cameras with wide-angle lenses are widely used in surveillance but often require image rectification due to their inherent nonlinear distortions. Current deep learning approaches typically struggle to maintain fine-grained geometric details, resulting in inaccurate rectification. This paper presents a Forward Distortion and Backward Warping Network (FDBW-Net), a novel framework for wide-angle image rectification. It begins by using a forward distortion model to synthesize barrel-distorted images, reducing pixel redundancy and preventing blur. The network employs a pyramid context encoder with attention mechanisms to generate backward warping flows containing geometric details. Then, a multi-scale decoder is used to restore distorted features and output rectified images. FDBW-Net's performance is validated on diverse datasets: public benchmarks, AirSim-rendered PTZ camera imagery, and real-scene PTZ camera datasets. It demonstrates that FDBW-Net achieves SOTA performance in distortion rectification, boosting the adaptability of PTZ cameras for practical visual applications.