RAW Image Encoding for Dual Bayer HDR Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for encoding RAW image data from color filter arrays, such as the Bayer arrangement, do not efficiently handle sensors with additional white pixels that enhance sensitivity at low illuminance, as they are not compatible with the plane conversion methods used in these arrays.
Innovation Solution
An image encoding apparatus that converts RAW image data from a color filter array with white pixels into multiple planes of single color components, using a plane conversion unit to generate low-frequency and high-frequency component planes, and then encodes these using a frequency transform and quantization process, allowing for efficient coding of the data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If plane conversion methods from Document 1 and Document 2 are used for Bayer arrangement, then coding efficiency is improved, but compatibility with dual Bayer+HDR array structure is lost
Solution Approach 1:
The patent segments the dual Bayer+HDR array data into five separate planes (R, G0, B, G1, and W) for independent processing. This segmentation allows each plane to be handled according to its specific characteristics, enabling efficient coding while maintaining compatibility with the unique sensor structure that includes white pixels for high dynamic range capture.
Solution Approach 2:
The patent creates a universal plane conversion method that works for both traditional Bayer arrangements and dual Bayer+HDR array structures. By generalizing the conversion approach to handle five color components (including white pixels), the method achieves multi-functionality that adapts to different sensor types while maintaining coding efficiency.
2Measurement precision
If RAW data is converted to RGB or YUV signal through demosaicing, then image quality is improved, but data amount increases to 3 times the RAW data
Solution Approach 1:
The patent performs plane conversion and frequency transform operations on RAW data before final encoding, preparing the data in an optimized format that maintains quality while controlling data volume. This preliminary processing allows subsequent encoding steps to be more efficient, avoiding the need to fully demosaic to RGB/YUV before compression.
Solution Approach 2:
The patent transforms the problem from color space conversion (RAW to RGB/YUV) to frequency domain transformation. By applying frequency transforms to the five color planes, the method separates frequency components that can be independently encoded, achieving compression without requiring full demosaicing to three-color-per-pixel format.
3Illumination intensity
If white pixels are added to increase sensitivity at low illuminance, then sensitivity is improved, but complexity of plane conversion method increases
Solution Approach 1:
The patent handles the five-color-component data by segmenting it into separate planes (R, G0, B, G1, W) rather than attempting complex interleaved processing. This segmentation simplifies the conversion methodology by treating each color component independently, making the system manageable despite the increased number of pixels and components.
Solution Approach 2:
The patent develops a unified plane conversion framework that handles both traditional three-color Bayer data and five-color dual Bayer+HDR data through the same general methodology. This universal approach avoids creating entirely separate complex processing paths, reducing overall system complexity while accommodating the additional white pixels.
Data Source
AI summary
This invention provides an image encoding apparatus operable to encode RAW image data obtained from a capturing sensor that has a color filter array in which a plurality of filters for each of three primary colors and a plurality of filters for a specific color for luminance are arranged in an N×N pixel region, where the image encoding apparatus comprises a conversion unit which converts the RAW image data into a plurality of planes each configured by a single color component; and an encoding unit which encodes each of the planes, wherein the conversion unit, for each component representing the three primary colors, by referencing pixel values of the same component in the N×N pixel region, generates a plane configured by a low-frequency component data and a plane configured by a high-frequency component.