Imaging Apparatus RAW Data Encoding with Luminance-Based Quantization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging apparatuses face challenges in encoding RAW image data with high efficiency while minimizing image quality degradation, particularly due to data compression, which can be exacerbated by varying shooting conditions and the need to suppress data volume in limited storage media.
Innovation Solution
An imaging apparatus that converts RAW image data from a Bayer array into multiple planes, performs frequency transformation to generate subbands, and applies non-linear conversion to coefficient data based on camera information to optimize quantization, especially in dark portions, thereby reducing quantization loss and maintaining image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the RAW image is compressed to suppress data amount, then storage space efficiency is improved, but image quality degradation occurs
Solution Approach 1:
The patent applies different quantization strengths to different regions of the image based on luminance levels. Dark portions (low luminance) are quantized with higher precision to preserve shadow details, while bright portions are quantized with lower precision. This local differentiation allows effective data compression while minimizing perceptible image quality degradation, as the human visual system is more sensitive to errors in dark regions.
Solution Approach 2:
The patent dynamically changes quantization parameters based on shooting conditions and image characteristics. The quantization step size is adjusted according to luminance values, with smaller steps for dark portions and larger steps for bright portions. This parameter adaptation enables the encoding system to optimize the balance between compression ratio and image quality for each specific image.
2Productivity
If quantization is changed according to human visual system characteristics, then encoding efficiency is improved, but image quality degradation occurs when developing process changes luminance and contrast
Solution Approach 1:
The patent performs non-linear conversion on the coefficient data before quantization to anticipate and compensate for subsequent developing process operations. By pre-adjusting the coefficient values based on expected luminance and contrast changes during development, the encoding ensures that image quality is preserved even after the developing process modifies the image data. This preliminary compensation prevents the quality degradation that would otherwise occur.
Solution Approach 2:
The patent incorporates feedback from camera information related to imaging conditions into the non-linear conversion process. The encoding system uses shooting parameters and image characteristics to dynamically adjust the non-linear conversion applied to coefficient data, creating a closed-loop system that adapts to specific imaging scenarios and maintains optimal image quality throughout the processing chain.
3Device complexity
If uniform quantization is applied to all coefficient data, then encoding process is simplified, but image quality degradation occurs in dark portions
Solution Approach 1:
The patent implements local quality differentiation by applying non-linear conversion specifically to coefficient data based on luminance characteristics. Instead of uniform quantization, the system identifies dark portions through luminance analysis and applies enhanced non-linear conversion to preserve shadow details. This targeted approach maintains image quality in critical dark regions while keeping the overall encoding process relatively simple through automated luminance-based classification.
Data Source
AI summary
This invention performs encoding with high encoding efficiency while suppressing image quality degradation when RAW image data obtained by imaging is to be encoded. Hence, an imaging apparatus that encodes an image captured by an image sensor includes a plane converting unit configured to convert RAW image data of a Bayer array obtained by the image sensor into a plurality of planes each comprised of a pixel of a single component, a frequency transforming unit configured to generate a plurality of subbands by frequency-transforming each of the obtained planes, a non-linear converting unit configured to non-linearly convert coefficient data forming each of the subbands based on camera information related to imaging in the imaging apparatus, and an encoding unit configured to encode by quantizing the coefficient data obtained by the non-linear conversion by the non-linear converting unit.