Depth-Assisted Perspective Distortion Correction for Close-Range Portraits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Close-range portraiture photographs taken with mobile device cameras often exhibit perspective distortions due to the mismatch between the camera's field of view and the viewing display configuration, which can magnify facial features like the nose and chin.
Innovation Solution
The system automatically corrects these distortions by using depth maps to detect and segment faces from the background, re-rendering the face from a more distant viewpoint, and combining it with the background to create a new image with reduced perspective distortion, employing techniques such as alpha matting, 3D warping, and inpainting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a wide angular field of view camera is used for close-range portraiture, then more of the scene can be captured, but perspective distortion magnifies facial features like nose and chin
Solution Approach 1:
The image is segmented into foreground (face) and background regions using depth map data. The face region is identified and separated from the rest of the image, allowing independent processing of the face portion to correct perspective distortion while preserving the original wide-field-of-view background.
Solution Approach 2:
The patent uses depth map data (adding a depth dimension) to perform 3D warping of the face region. By introducing depth information, the system can re-render the face from a synthesized viewpoint that corresponds to a more distant position, effectively correcting perspective distortion in the 2D image plane.
2Length of stationary object
If the camera is positioned close to the subject for portraiture, then intimate detail is captured, but perspective distortion increases
Solution Approach 1:
The patent creates a synthetic copy of the face at a different virtual distance using depth warping. Instead of physically moving the camera, the system generates a warped version of the face that appears as if captured from a more distant viewpoint, then composites this warped copy back into the original image.
Solution Approach 2:
Depth map data provides a third dimension (distance from camera) that enables the system to calculate and apply perspective corrections. By using the depth information, the system can simulate what the face would look like from a different distance without actually changing the camera position.
3Shape
If depth map data is used to segment and rerender faces, then perspective distortion is corrected, but processing complexity increases
Solution Approach 1:
The depth map is captured simultaneously with the color image, providing pre-computed depth information that simplifies subsequent processing. By having depth data available beforehand, the system avoids the need for complex post-processing depth estimation algorithms.
Solution Approach 2:
The patent applies complex warping and compositing operations only to the face region (foreground) rather than the entire image. This localized processing approach reduces overall computational complexity while still achieving the desired perspective correction where it matters most.
Data Source
AI summary
Systems and methods for automatically correcting apparent distortions in close range photographs that are captured using an imaging system capable of capturing images and depth maps are disclosed. In many embodiments, faces are automatically detected and segmented from images using a depth-assisted alpha matting. The detected faces can then be re-rendered from a more distant viewpoint and composited with the background to create a new image in which apparent perspective distortion is reduced.