AI Image Acquisition for Sub-Object Removal and Restoration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image acquisition systems struggle to effectively distinguish and remove sub-objects from images, leading to incomplete representation of main objects hidden by these sub-objects, and lack the ability to restore hidden portions of main objects.
Innovation Solution
An image acquisition device utilizing AI neural networks to detect main and sub-objects, remove sub-objects, and restore hidden portions of main objects, with optional indicators and sharpness adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sub-objects are removed from the image, then the main object becomes more visible and recognizable, but the hidden portions of the main object cannot be restored
Solution Approach 1:
The patent introduces an AI neural network as an intermediary between the captured image and the final processed image. The neural network performs two key functions: (1) detects and removes sub-objects that obscure the main object, and (2) generates and fills in the hidden portions of the main object based on learned patterns and context. This intermediary system resolves the contradiction by enabling both sub-object removal and hidden portion restoration simultaneously.
Solution Approach 2:
The patent replaces traditional mechanical or rule-based image processing methods with an AI neural network system. Instead of using conventional algorithms that can only remove sub-objects but cannot reconstruct hidden areas, the neural network uses deep learning to both remove obstructions and intelligently generate missing object portions, substituting the limited mechanical approach with a more advanced intelligent system.
2Measurement precision
If AI neural networks are used to detect and remove sub-objects, then image processing capability is improved, but device complexity increases
Solution Approach 1:
The patent implements a multi-functional AI neural network system that performs multiple operations within a single integrated framework: (1) detecting the main object, (2) detecting sub-objects, (3) removing sub-objects, and (4) restoring hidden portions. By consolidating these functions into a universal neural network system, the patent manages complexity while achieving high detection and processing accuracy.
3Manufacturing precision
If hidden portions of the main object are restored using AI, then complete representation is achieved, but processing time increases
Solution Approach 1:
The patent employs a neural network that has been pre-trained on large datasets of objects and scenes. This preliminary training allows the network to have prior knowledge about object structures, patterns, and contexts. When processing a new image, the pre-trained network can quickly generate and fill hidden portions without requiring extensive computation, thereby reducing processing time while maintaining high restoration quality.
Data Source
AI summary
Provided is an artificial intelligence (AI) system that mimics functions, such as recognition and determination by human brains, by utilizing a machine learning algorithm, such as deep learning, and applications of the AI system. An image acquisition device is disclosed including a camera configured to acquire a first image, wherein a portion of a main object is hidden from the camera by a sub-object; at least one processor configured to input the first image to a first AI neural network; detect, by the first AI neural network from data corresponding to a plurality of objects included in the first image, first data corresponding to the main object and second data corresponding to the sub-object from the first image by inputting the first image to an AI neural network, remove the sub-object from the first image, and generate, using a second AI neural network, a second image by restoring third data corresponding to at least a portion of the main object hidden by the removed sub-object by using the AI neural network, wherein the third data replaces the second data; and a display configured to display at least one of the first image and the second image.