AI Local Dimming Control for Backlight Contrast and Power Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing display technologies, such as rule-based systems, fail to effectively implement local dimming methods that consider the features of output images, resulting in suboptimal contrast ratios and power consumption in non-self-luminous display devices.
Innovation Solution
A display apparatus utilizing an artificial intelligence (AI) model to determine the driving current for backlight blocks based on pixel information from input images, trained on luminance data from sample images, to perform local dimming and enhance contrast ratio.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If rule-based local dimming is used to control backlight blocks, then power consumption is reduced, but contrast ratio deteriorates because the method does not consider output image features
Solution Approach 1:
The system uses an AI model that takes both input image pixel values and output image features as inputs to determine backlight block duties. This feedback mechanism allows the system to adjust backlight intensity based on the actual output image characteristics, thereby improving contrast ratio while maintaining power efficiency. The AI model learns from training data the optimal relationship between input images, output features, and backlight control parameters.
Solution Approach 2:
The system dynamically changes the duty parameters of backlight blocks based on analyzed image features. By adjusting the duty cycle of backlight blocks according to local brightness requirements derived from AI analysis, the system optimizes both power consumption and contrast ratio. The AI model determines optimal duty values that balance energy efficiency with image quality requirements.
2Manufacturing precision
If AI model is applied to determine backlight block duties based on pixel information, then contrast ratio is improved, but device complexity increases
Solution Approach 1:
The AI model serves as an intermediary between the input image data and the backlight control system. Instead of directly implementing complex control logic in the display driver, the system uses the trained AI model to process image data and generate backlight duty parameters. This intermediary approach simplifies the overall system architecture while achieving sophisticated contrast optimization through the AI's pattern recognition capabilities.
Solution Approach 2:
The AI model is trained in advance using training images and their corresponding ground truth data to learn the optimal mapping from image features to backlight duties. This preliminary training action allows the model to be deployed with pre-learned knowledge, reducing the computational complexity during actual operation. The system benefits from the AI model's pre-acquired understanding of image-backlight relationships without requiring complex real-time calculations.
3Device complexity
If existing rule-based methods are used to control backlight, then device complexity is kept low, but image quality and power efficiency deteriorate
Solution Approach 1:
The AI model enables the display system to automatically optimize its own backlight control based on image content analysis. The system self-adjusts backlight block duties by analyzing pixel information and output image features without requiring manual intervention or complex external control systems. This self-service capability achieves improved power efficiency while maintaining relatively simple device architecture through the AI model's autonomous decision-making.
Data Source
AI summary
A display apparatus includes: a display panel; a backlight unit including a plurality of backlight blocks; and a processor obtaining a current duty of a driving current for driving each of the plurality of backlight blocks by applying an artificial intelligence (AI) model to pixel information of an input image and driving the backlight unit based on the obtained current duty, in which the AI model is a model trained based on first luminance information included in an output image corresponding to each of a plurality of sample images and second luminance information corresponding to pixel information included in each of the plurality of sample images.