Image Processing for Authentic Moving Object Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to conveniently and authentically remove moving objects from single photos captured in a default mode, often resulting in inaccurately filled regions due to guessed image content.
Innovation Solution
An image processing method that groups similar images based on time and location, allowing for the removal of moving objects by replacing them with authentic content from other images, enhancing the authenticity and convenience of the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If AI guessing technology is used to fill regions after eliminating a person from a single photo, then the elimination function can be provided, but the authenticity of the filled content is low and greatly differs from the surrounding environment
Solution Approach 1:
The patent uses actual content from other captured images to copy and fill the region where the moving object is eliminated, rather than using AI-generated guessed content. This ensures the filled content is authentic and consistent with the real surrounding environment, directly resolving the authenticity problem while maintaining operational convenience.
Solution Approach 2:
The system captures multiple images in advance before the moving object elimination is needed. These pre-captured images serve as a pool of authentic content that can be used to fill eliminated regions, ensuring both convenience and high authenticity without requiring complex real-time AI generation.
2Reliability
If multiple frames of images are captured in a specific photographing mode to eliminate passersby, then the elimination function works, but the operation becomes inconvenient and cannot process single photos in default mode
Solution Approach 1:
The patent makes the moving object elimination function universal by enabling it to work with single photos captured in the default photographing mode, rather than requiring a specific multi-frame capture mode. The system can process both single images and multiple images, making the function adaptable to various user scenarios and significantly improving ease of operation.
Solution Approach 2:
The system automatically selects and processes images from the captured set without requiring manual user intervention to choose specific frames. The electronic device itself performs the selection and elimination process, making the operation self-service and convenient for users who simply need to capture a photo and have the moving object removed.
Data Source
AI summary
Embodiments of this application provide an image processing method. The method includes: A plurality of images that meet a first condition are determined, where the first condition includes that a similarity between any two images in the plurality of images is greater than or equal to a first threshold. A first image that meets a second condition is determined from the plurality of images, where the first image includes a first object, and the first object is a to-be-eliminated object in the first image. A second image is determined from the plurality of images, where the second image includes a second object, and a position of the second object in the second image corresponds to a position of the first object in the first image. The first object is covered by or replaced with the second object, to obtain a target image.