Oblique Image Stitching via 3D Geometric Proxy and Seam Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image stitching techniques fail to produce photorealistic mosaic images from oblique views, especially when capturing tall buildings from different angles, resulting in unnatural angles and warping due to misaligned textures and shadows.
Innovation Solution
The Oblique Image Stitcher constructs a photorealistic mosaic image by projecting input images onto a rough 3D model using projective texture mapping, identifying optimal seams with a modified graph-cut analysis, and blending images using conventional techniques to minimize artifacts and ensure photorealism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional image stitching techniques are used to create mosaic images from oblique views, then the stitching process can be completed, but the resulting images contain unnatural angles and warping due to misaligned textures and shadows
Solution Approach 1:
The patent transitions from 2D image stitching to 3D geometric modeling by introducing a rough 3D model of the scene. Input images are projected onto this 3D model, allowing the system to account for the actual spatial geometry of buildings and terrain, thereby resolving alignment issues that cause unnatural angles and warping in traditional 2D stitching approaches.
Solution Approach 2:
The patent introduces a rough 3D model as an intermediary between the input oblique images and the final mosaic output. This 3D model serves as a common geometric framework that mediates the alignment and integration of multiple images, enabling photorealistic reconstruction by properly positioning textures and shadows in three-dimensional space before projecting the final view.
2Measurement precision
If LiDAR-based techniques are used to create detailed 3D models, then resolution can be improved, but the models become noisy and require computationally expensive smoothing
Solution Approach 1:
The patent deliberately uses a rough, lower-resolution 3D model rather than attempting to create a perfectly detailed model. This partial approach avoids the noisy artifacts and computational complexity of high-precision LiDAR processing while still providing sufficient geometric accuracy for the intended application of oblique mosaic generation.
Solution Approach 2:
The patent extracts only the essential geometric information needed for oblique view reconstruction from the 3D model, rather than processing the complete high-fidelity model. By taking out only the necessary geometric framework and ignoring excessive detail, the system achieves good results without the computational burden of smoothing noisy high-resolution data.
3Productivity
If conventional mosaicing algorithms are used to align images, then the stitching can be performed, but occlusion boundaries of buildings are misaligned resulting in mismatched textures and shadows
Solution Approach 1:
The patent resolves occlusion boundary misalignment by working in 3D space rather than 2D image space. By projecting images onto a 3D model and rendering from the desired viewpoint, the system naturally handles occlusions according to the actual spatial relationships of buildings, ensuring that textures and shadows align correctly at occlusion boundaries without requiring complex 2D image manipulation.
4Area of stationary object
If multiple oblique images captured from different angles are stitched together, then coverage of the scene can be increased, but buildings appear to lean at different angles creating a confusing jumble
Solution Approach 1:
The patent achieves both wide scene coverage and geometric consistency by constructing a 3D model from multiple oblique images and then rendering a unified oblique view from a fixed viewpoint. This allows the system to incorporate data from images taken at various angles and positions while presenting a consistent geometric composition, as all images are re-projected to match the target viewpoint rather than being stitched with their original orientations preserved.
Data Source
AI summary
An “Oblique Image Stitcher” provides a technique for constructing a photorealistic oblique view from a set of input images representing a series of partially overlapping views of a scene. The Oblique Image Stitcher first projects each input image onto a geometric proxy of the scene and renders the images from a desired viewpoint. Once the images have been projected onto the geometric proxy, the rendered images are evaluated to identify optimum seams along which the various images are to be blended. Once the optimum seams are selected, the images are remapped relative to those seams by leaving the mapping unchanged at the seams and interpolating a smooth mapping between the seams. The remapped images are then composited to construct the final mosaiced oblique view of the scene. The result is a mosaic image constructed by warping the input images in a photorealistic manner which agrees at seams between images.