Backlight-Scaled Image Enhancement via Luminance Decomposition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional methods fail to effectively enhance images with very dim backlight, such as 10% or 5% of full backlight, leading to detail loss and color degradation, which compromises battery life without maintaining image quality.

Innovation Solution

A method that boosts luminance of dark image areas below a perceptual threshold while preserving contrast, using a human visual system (HVS) response model to decompose the luminance layer into an HVS response layer and a background luminance layer, and then enhancing the background luminance layer to a perceptible range, avoiding luminance gradient reversal and over-compensation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by moving object

If LCD backlight is reduced to extend battery life, then energy consumption is reduced, but image quality deteriorates due to detail loss and color degradation

Engineering Contradiction:

Improvebacklight energy consumptionVSAvoidimage detail visibility

Core Design Contradiction:

Use of energy by moving objectVSLoss of information

Solution Approach 1:

The image is segmented into multiple layers including luminance layer, chrominance layer, and detail layer. The luminance layer is further decomposed into foreground luminance map and background luminance map, allowing selective enhancement of dark regions without affecting the entire image or other color components.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different processing strategies are applied to different regions of the image. Dark regions below the perceptual threshold receive luminance boosting, while bright regions maintain their original characteristics. This localized processing preserves image quality in critical areas while minimizing overall energy consumption.

Inventive Principle:

Principle #3Local quality

2Loss of information

If conventional enhancement methods are applied to dim backlight images, then some image quality improvement is achieved, but luminance gradient reversal and over-compensation occur

Engineering Contradiction:

Improveimage detail visibilityVSAvoidluminance gradient accuracy

Core Design Contradiction:

Loss of informationVSReliability

Solution Approach 1:

The image is pre-processed by decomposing the luminance layer into foreground and background components before enhancement. The foreground luminance map is extracted and preserved, while only the background luminance map undergoes boosting. This preliminary separation prevents gradient reversal by ensuring edge information is not corrupted during enhancement.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The perceptual threshold acts as an intermediary criterion to control the enhancement process. By comparing pixel luminance values against this threshold, the system selectively boosts only those regions that benefit from enhancement, avoiding over-compensation in already visible regions and preventing artifacts.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS9367905B2Method and system of enhancing a backlight-scaled image

Publication Date: 2016.06.14 HIMAX TECH LTD

AI summary

A method and system of enhancing a backlight-scaled image include a minimum perceptible luminance threshold of cone response with dim backlight being determined, and a luminance layer associated with an image being extracted. The luminance layer is decomposed into an HVS response layer and a background luminance layer for each pixel of the luminance layer. Luminance of dark pixels of the background luminance layer is boosted and compressed to a perceptible range above the minimum perceptible luminance threshold, thereby resulting in an enhanced background luminance layer. An enhanced luminance layer is generated through composition using the HVS response layer and the enhanced background luminance layer as inputs.