Dynamic Tone Mapping with Layered 2D LUTs for Image Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional color and tone mapping methods in digital cameras often result in undesirable artifacts and suboptimal image quality, particularly when capturing scenes with both bright and dark features due to limited bit depth and inadequate dynamic range.
Innovation Solution
A camera system equipped with an image signal processor (ISP) that generates 2D color look-up tables (LUTs) for each Y-layer of YCbCr image data, converting raw RGB data into optimized CbCr data and tone-mapping it to produce optimized sRGB images, using techniques like bilinear interpolation and chroma gain maps to enhance color and tone representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional color mapping methods adjust chroma and hue using 3D LUTs, then color reproduction is achieved, but smoothness for various hues with varying chroma deteriorates
Solution Approach 1:
The patent segments the color mapping process by separating chroma adjustment from hue adjustment, and further divides the image into multiple Y-layers based on luminance ranges. Instead of applying a single 3D LUT to the entire image, the system generates separate 2D LUTs for each Y-layer, allowing independent optimization of color and tone mapping for different luminance regions, thereby improving hue smoothness while maintaining color accuracy
Solution Approach 2:
The patent transitions from conventional 3D LUTs (covering chroma, hue, and lightness) to 2D LUTs that operate on specific Y-layers. By adding the luminance-based layering dimension and applying 2D LUTs selectively to each layer, the system achieves more precise control over color and tone mapping, resolving the trade-off between color accuracy and hue smoothness
2Device complexity
If a single tone map is applied to images with both shadowed and bright features, then processing simplicity is maintained, but features on either edge of the luminance spectrum are compressed due to limited bit depth
Solution Approach 1:
The patent divides the image into multiple Y-layers based on luminance ranges, with each layer receiving a dedicated tone map optimized for its specific luminance characteristics. This segmentation allows shadowed regions, mid-tone regions, and bright regions to each have appropriate bit value allocation, preventing compression of features at luminance extremes while maintaining manageable processing complexity through systematic layer-based treatment
Solution Approach 2:
The patent applies different tone mapping characteristics to different luminance regions by generating specific tone maps for each Y-layer. Each layer receives a tone map with bit values allocated according to its luminance characteristics, ensuring that shadowed regions get more bits for dark features while bright regions get appropriate allocation for highlights, achieving local optimization of luminance representation
3Ease of operation
If general tone maps allocate bit values evenly across the luminance spectrum, then implementation simplicity is achieved, but adaptation to specific image content deteriorates
Solution Approach 1:
The patent implements dynamic tone mapping by generating multiple tone maps corresponding to different Y-layers based on the actual luminance distribution in the image. Rather than using a single static tone map, the system adaptively creates layer-specific tone maps that respond to the image content, allowing even implementation within each layer while achieving overall adaptation to the specific scene being captured
Data Source
AI summary
The processing of RGB image data can be optimized by performing optimization operations on the image data when it is converted into the YCbCr color space. For each Y-layer of the YCbCr image data, a 2D LUT is generated. The YCbCr image data is converted into optimized CbCr image data using the 2D LUTs, and optimized YCbCr image data is generated by blending CbCr image data corresponding to multiple Y-layers. The optimized YCbCr image data is converted into sRGB image data, and a tone curve is applied to the sRGB image data to produce optimized sRGB image data.