Chromatically Subsampled Image Dithering via HVS Optical Transfer Function
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image dithering techniques face limitations in preserving high dynamic range (HDR) and visual dynamic range (VDR) images due to hardware constraints in display processing, which restrict bit depth precision and result in suboptimal noise visibility and bit depth reduction.
Innovation Solution
The method involves using a model of the human visual system's optical transfer function (OTF) to shape noise added to chromatically subsampled images, ensuring the noise is spectrally inverted and equally invisible across frequencies, primarily by adding noise to chrominance components and compensating for display and viewing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dithering is applied to reduce bit-depth in chromatically subsampled images, then bit depth precision is improved, but noise visibility deteriorates
Solution Approach 1:
The patent applies different dithering strategies to different color components (luma vs. chroma) based on their respective sensitivities to noise. The chroma components receive dithering treatment optimized for their lower visual sensitivity, while luma components receive different treatment, thereby achieving local optimization of noise visibility across the image spectrum.
Solution Approach 2:
The patent modifies the dithering parameters specifically for chromatically subsampled images by incorporating the optical transfer function (OTF) model to adjust noise characteristics. This involves changing the spectral distribution and amplitude of dither noise to match the chroma subsampling characteristics, thereby optimizing the trade-off between bit-depth reduction and noise visibility.
2Device complexity
If hardware constraints limit processing precision, then device complexity is reduced, but manufacturing precision deteriorates
Solution Approach 1:
The patent applies dithering noise addition before the quantization step, which is a preliminary action that prepares the image data for subsequent bit-depth reduction. By pre-processing the image with optimized dithering patterns that account for chroma subsampling, the system achieves better final precision despite hardware limitations in the quantization stage.
Solution Approach 2:
The patent introduces dithering noise as an intermediary element between the high-precision input image and the low-precision output. This intermediary noise pattern acts as a mediator that distributes quantization errors in a visually acceptable manner, particularly optimized for chromatically subsampled formats where chroma noise is less perceptible.
3Productivity
If noise is added to dither images, then bit depth reduction is improved, but noise invisibility deteriorates
Solution Approach 1:
The patent inverts the conventional approach by designing dither noise patterns that are specifically tailored to be invisible in chromatically subsampled images. Instead of using generic dither patterns and hoping for the best, the system inverts the problem by first analyzing the visual sensitivity characteristics of subsampled chroma and then generating noise patterns that exploit this insensitivity, making the noise invisible rather than trying to hide it.
Solution Approach 2:
The patent changes the spectral and spatial parameters of dither noise to optimize for chroma subsampling. By adjusting the noise frequency content, amplitude distribution, and spatial correlation based on the OTF model and chroma subsampling characteristics, the system achieves effective bit-depth reduction while maintaining noise invisibility in the chroma channels.
Data Source
Figure 1A~1B
Figure 2A
Figure 2B
AI summary
Dithering techniques for images are described herein. An input image of a first bit depth is separated into a luma and one or more chroma components. A model of the optical transfer function (OTF) of the human visual system (HVS) is used to generate dither noise which is added to the chroma components of the input image. The model of the OTF is adapted in response to viewing distances determined based on the spatial resolution of the chroma components. An image based on the original input luma component and the noise-modified chroma components is quantized to a second bit depth, which is lower than the first bit depth, to generate an output dithered image.