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.

A Stitch Partition Algorithm Based on BP Neural Network

THE RESEARCH OF ALGORITHM OF STROKE BASED ON B-SPLINE IN FABRIC CAD SYSTEM

RecStitchNet: Learning to Stitch Images with Rectangular Boundaries

FSIS: Fast and Seamless Image Stitching of High-Resolution Dunhuang Murals

Image Stitching and Blending of Dunhuang Murals Based on Image Pyramid

Stitching Patch Arrangement Planning for Simulation of Irregular Needling Embroidery

Research on Approach of Generating Random Stitch in Embroidery CAD

Studies on Image Stitching Algorithms in Machine Vision Inspection of Solar Panel

Shape Decomposition Algorithm in Embroidery

DunHuangStitch: Unsupervised Deep Image Stitching of Dunhuang Murals

StitchGeneration: Modeling and Creation of Random-Needle Embroidery Based on Markov Chain Model

A Controllable Stitch Layout Strategy for Random Needle Embroidery

Multilayered Stitch Generating for Random-Needle Embroidery

Design of Testing-and-evaluating System for Sewing Performance of Sewing Machines

StitchNet: Composing Neural Networks from Pre-Trained Fragments

A SELF-ORGANIZING NEURAL NETWORK FOR PARTITIONING ON MCM

Stitches Generation for Random-Needle Embroidery Based on Markov Chain Model.

Towards End-to-End Embroidery Style Generation: A Paired Dataset and Benchmark

Segmentation and Synthesis of Embroidery Art Images Based on Deep Learning Convolutional Neural Networks

Stitch-Based Image Stylization for Thread Art Using Sparse Modeling.

Vector and Pixel Space Depended Stitch Definition and Style Transfer for Random-Needle Embroidery