Multiple Image Blender for HDR Exposure Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image blending methods for high dynamic range (HDR) images deteriorate picture quality when applied to displays with small dynamic ranges, as they fail to effectively combine images captured with different exposure settings.
Innovation Solution
A multiple image blender that includes a controller and an image processor to blend HDR images or multiple images with different exposures, using weight generation and area/gradation level setting to optimize image combination, ensuring overlapping dynamic ranges and optimal gradation levels, thereby generating high-quality multi-exposure images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional image blending methods are used to combine HDR images with different exposure settings, then the dynamic range of the image is expanded, but the picture quality deteriorates when displayed on devices with small dynamic ranges
Solution Approach 1:
The image is divided into multiple regions based on luminance levels (bright, middle, dark regions). Different blending strategies are applied to each region: bright regions use short-exposure images, dark regions use long-exposure images, and middle regions use a combination. This segmentation allows the blending process to preserve picture quality in each specific luminance range while maintaining overall dynamic range expansion.
Solution Approach 2:
Different quality parameters are optimized for different luminance regions. The blending weights and processing parameters are locally adjusted based on the luminance characteristics of each region, ensuring that picture quality is maintained in display-relevant regions while still achieving dynamic range expansion across the full image.
2Ease of operation
If simple average luminance blending is used, then the processing is simple and fast, but the picture quality deteriorates due to loss of detail in varying luminance regions
Solution Approach 1:
The image processing is segmented into multiple stages: region classification based on luminance, weight generation for each region, and selective blending. This structured approach maintains processing efficiency while significantly improving picture quality by preserving details in different luminance regions through region-specific blending parameters.
Solution Approach 2:
The blending parameters (weights, exposure compensation values) are dynamically changed based on the luminance characteristics of different image regions. This allows the processing to adapt to local image characteristics, preserving picture quality while maintaining computational efficiency through parameter-based control rather than complex algorithms.
3Device complexity
If uniform blending weights are applied across the entire image, then the processing is straightforward, but the picture quality deteriorates due to loss of detail in specific luminance regions
Solution Approach 1:
Uniform blending weights are replaced with region-specific weights that vary according to luminance levels. Bright regions receive weights favoring short-exposure images, dark regions receive weights favoring long-exposure images, and middle regions use balanced weights. This local quality approach significantly improves picture quality by preserving details in each luminance region while adding only moderate complexity through region classification.
Solution Approach 2:
The blending weights are changed as a function of luminance level, creating a smooth transition between different exposure contributions across the image. This parameter-based approach improves picture quality by adapting to local luminance characteristics while maintaining relatively simple processing through continuous weight functions rather than complex region-based algorithms.
Data Source
AI summary
Disclosed is an apparatus to blend a high dynamic range (HDR) image or a plurality of images captured with different exposure settings to multiple images, and a method thereof. A multiple image blender receives a high dynamic range (HDR) image or a plurality of images captured with different exposure settings, and controls setting of one or more areas of interest or setting of one or more gradation levels, and combines and blends the HDR image or the plurality of images captured with different exposure settings corresponding to each area of interest or at each gradation level, and generates at least one multi-exposure image.