.png){kind=icon}
返回
Efficient Superpixel-Based Seamline Detection for Large-Scale Image Stitching
.png){kind=icon}
PRE
AI
DOI：10.1109/LGRS.2025.3548266.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
As a crucial procedure for image stitching, seamline detection has an important impact on the quality of the final mosaic. However, traditional seamline detection methods are less efficient in determining the optimal seamlines for multiple images. This letter presents an efficient superpixel-level seamline detection method for large-scale unmanned aerial vehicle (UAV) image stitching, which can detect the optimal seamlines for hundreds of images in several minutes. Specifically, a novel superpixel-based energy function which simultaneously considers color difference, gradient magnitude, and texture complexity is devised to determine the superpixel-level optimal seamlines in the overlapping areas of multiple images. The energy minimization problem is efficiently solved by the multilabel optimization algorithm. Experimental results on five remote sensing datasets captured by UAVs have shown that the proposed superpixel-level seamline detection method is very effective in generating high-quality seamless mosaics for large-scale images.
Keyword:
Optimization
Image stitching
Image segmentation
Image color analysis
Complexity theory
Minimization
Autonomous aerial vehicles
Remote sensing
Labeling
Costs
Multilabel optimization
multiple image stitching
seamline detection
superpixel segmentation
unmanned aerial vehicle (UAV) image
