Dual-Lens Camera Live View Correction for Screen Type Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic apparatuses struggle to accurately determine the type of screen in an image due to limited angle of view and difficulty in identifying the overall configuration, often providing inappropriate filters or failing to provide any filters, leading to misrecognition.
Innovation Solution
The electronic apparatus employs multiple lenses with different angles of view, using neural network models to analyze depth maps, saliency information, and object relationships to identify the screen type accurately and apply corresponding filters for image correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single lens is used to capture the image, then the device complexity is reduced, but the measurement precision of screen type identification deteriorates due to limited angle of view
Solution Approach 1:
The camera is divided into multiple lenses with different angles of view (first lens with narrow angle, second lens with wide angle). Each lens captures different portions of the scene, allowing the system to segment the viewing field and combine information from multiple perspectives to accurately identify screen types while maintaining reasonable device complexity.
2Measurement precision
If multiple neural network models are used to analyze image information, then the screen type identification accuracy is improved, but the processing time and computational complexity increase
Solution Approach 1:
The system performs preliminary actions by first capturing images with multiple lenses, then extracting key features (depth information, saliency maps, object relationships) before applying neural network models for screen type identification. This structured preliminary processing organizes data in advance, making the subsequent neural network analysis more efficient and accurate.
Solution Approach 2:
The image analysis process is segmented into multiple specialized neural network models, each handling specific aspects: one model for depth information extraction, another for saliency detection, and a third for object relationship analysis. This segmentation allows parallel processing of different features, improving overall efficiency while maintaining high identification accuracy.
3Measurement precision
If the angle of view is limited to a narrow range, then the focus on specific objects is improved, but the overall screen configuration understanding deteriorates
Solution Approach 1:
The system merges images captured by the first lens (narrow angle, high object detail) and the second lens (wide angle, broad scene coverage). By combining these complementary views, the system achieves both precise object detection from the narrow-angle image and comprehensive screen configuration understanding from the wide-angle image, resolving the contradiction between focus and overall context.
Data Source
AI summary
An electronic apparatus includes a camera including a first lens and a second lens capable of obtaining an image having an angle of view different from the first lens, a display, a memory, and a processor. To perform a controlling method, the processor is configured to provide a first image obtained using the first lens to the display as a live view, obtain a second image using the second lens, obtain image information regarding the second image and object information regarding at least one object depicted in the second image using at least one neural network model, identify a screen type of the second image based on the image information and the object information, identify a set of filters corresponding to the screen type of the second image, and correct the first image based on the identified set of filters to provide a corrected first image as the live view.