Abstract:Irregular boundaries in image stitching naturally occur due to freely moving cameras. To deal with this problem, existing methods focus on optimizing mesh warping to make boundaries regular using the traditional explicit solution. However, previous methods always depend on hand-crafted features (e.g., keypoints and line segments). Thus, failures often happen in overlapping regions without distinctive features. In this paper, we address this problem by proposing RecStitchNet, a reasonable and effective network for image stitching with rectangular boundaries. Considering that both stitching and imposing rectangularity are non-trivial tasks in the learning-based framework, we propose a three-step progressive learning based strategy, which not only simplifies this task, but gradually achieves a good balance between stitching and imposing rectangularity. In the first step, we perform initial stitching by a pre-trained state-of-the-art image stitching model, to produce initially warped stitching results without considering the boundary constraint. Then, we use a regression network with a comprehensive objective regarding mesh, perception, and shape to further encourage the stitched meshes to have rectangular boundaries with high content fidelity. Finally, we propose an unsupervised instance-wise optimization strategy to refine the stitched meshes iteratively, which can effectively improve the stitching results in terms of feature alignment, as well as boundary and structure preservation. Due to the lack of stitching datasets and the difficulty of label generation, we propose to generate a stitching dataset with rectangular stitched images as pseudo-ground-truth labels, and the performance upper bound induced from the it can be broken by our unsupervised refinement. Qualitative and quantitative results and evaluations demonstrate the advantages of our method over the state-of-the-art.

Content-Preserving Image Stitching With Piecewise Rectangular Boundary Constraints

Content-Preserving Image Stitching with Regular Boundary Constraints

RecStitchNet: Learning to Stitch Images with Rectangular Boundaries

Remote Sensing Image Stitch Using Modified Structure Deformation

A Robust Image Stitching Based on Solution Set Smoothing

Stitching Videos From Unstructured Camera Arrays With Rectangular Boundaries

Multi-Viewpoint Panorama Construction with Wide-Baseline Images

Structure Preservation and Seam Optimization for Parallax-Tolerant Image Stitching

Multiple Combined Constraints for Image Stitching

Parallax-Tolerant Image Stitching with Epipolar Displacement Field

Research on Image Stitching Algorithm Based on Point-Line Consistency and Local Edge Feature Constraints

Natural Image Stitching With Layered Warping Constraint

An improved adaptive triangular mesh-based image warping method

A Natural Shape-Preserving Stereoscopic Image Stitching.

Ratio-Preserving Half-Cylindrical Warps for Natural Image Stitching

Image stitching by line-guided local warping with global similarity constraint

Semantic Aware Stitching for Panorama

SC-AOF: A Sliding Camera and Asymmetric Optical-Flow-Based Blending Method for Image Stitching

Automatic Stitching System for Images of Plane-Like Scenes

Image Stitching of Scenes with Large Misregistration

Deep Rectangling for Image Stitching: A Learning Baseline