VIFNet: An End-to-end Visible-Infrared Fusion Network for Robust Image Dehazing
The proposed VIFNet is an end-to-end multimodal fusion network that effectively combines visible and infrared modalities to restore high-quality haze-free images, outperforming state-of-the-art single-modality dehazing methods.
Rolling Shutter Distortion Correction via Direct Intermediate Distortion Flow Estimation
The paper proposes a novel framework that directly estimates the intermediate distortion flow from the underlying global shutter image to the rolling shutter image, enabling efficient and high-quality rolling shutter correction.
Automated Generation of Realistic Image Splicing Datasets Using Advanced Image Composition Techniques
The core message of this paper is to propose an automated framework for generating realistic image splicing datasets using state-of-the-art image composition techniques, in order to bridge the gap between the quality and quantity of existing image forgery datasets and the real-world manipulations.
Improving Semantic Correspondence for Small Objects through Keypoint Bounding Box-Centered Cropping
The core message of this paper is to introduce a novel problem called 'Small Object Semantic Correspondence (SOSC)' and propose a Keypoint Bounding box-centered Cropping (KBC) method to address the challenge of closely located keypoints associated with small objects, which leads to the fusion of their features and makes it difficult to identify the corresponding keypoints.
Preserving Object-Level Geometric Structures for Natural and Distortion-Free Image Stitching
The proposed OBJ-GSP algorithm leverages semantic segmentation to extract object-level geometric structures and preserves them during image stitching, achieving superior alignment and distortion prevention compared to existing methods.
© 2024 by Linnk AI