RAW Image Prediction Switching for 2×2 OCL Encoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The application of Multi-Pixel Compressor (MPC) encoding methods to 2×2 on-chip lens (OCL) RAW images risks reducing encoding efficiency due to pixel value deviations caused by varying light ray directions.
Innovation Solution
Implementing spatial and phase prediction methods to derive prediction values for pixels in a 2×2 OCL configuration, where spatial prediction refers to peripheral pixel values near the processed pixel and phase prediction refers to same-location pixel values within the block, to encode prediction residuals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If Multi-Pixel Compressor (MPC) encoding method is applied to 2×2 on-chip lens (OCL) RAW images, then compatibility with quad configuration is improved, but encoding efficiency deteriorates due to pixel value deviations caused by varying light ray directions
Solution Approach 1:
The patent applies different prediction methods to different pixel locations within the 2×2 block. Specifically, it uses phase prediction for pixels where the deviation pattern matches the phase prediction model, and spatial prediction for other cases. This localized adaptation of prediction strategies optimizes encoding efficiency for each pixel based on its specific characteristics within the block, resolving the contradiction between maintaining compatibility and preserving encoding efficiency.
Solution Approach 2:
The patent dynamically selects between phase prediction and spatial prediction methods based on the actual pixel value distribution and deviation patterns observed in the 2×2 block. Rather than applying a fixed prediction method, the system adapts its prediction approach according to the specific lighting conditions and pixel characteristics, thereby maintaining high encoding efficiency while remaining compatible with 2×2 OCL configurations.
2Measurement precision
If phase prediction is used for 2×2 OCL pixels, then prediction accuracy for same-location pixels is improved, but prediction accuracy deteriorates when pixel values deviate due to light ray direction variations
Solution Approach 1:
The patent incorporates a feedback mechanism that evaluates the actual pixel value distribution within the 2×2 block and uses this information to select the most appropriate prediction method. By monitoring the deviation patterns and comparing them against the expected patterns for phase prediction, the system can determine whether phase prediction or spatial prediction will yield better results, thereby maintaining both accuracy and reliability under varying light conditions.
Solution Approach 2:
The patent changes the prediction method parameter dynamically based on the observed pixel value characteristics. When pixel values show patterns consistent with phase prediction assumptions, the system uses phase prediction; when deviations indicate otherwise, it switches to spatial prediction. This parameter adaptation allows the system to maintain high prediction accuracy and reliability across different lighting scenarios.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach maintains encoding efficiency by applying the most accurate prediction method based on pixel value distribution characteristics, suppressing drops in encoding efficiency.
Implementation Method 1
Each of the plurality of pixels in the block has a filter having identical transmission wavelength characteristics, and is configured to photoelectrically convert incident light incident through a single on-chip lens corresponding to the plurality of pixels and the filter
Data Source
AI summary
Provided is an image processing device to suppress a drop in encoding efficiency. One of spatial prediction and phase prediction is set as a prediction method for a pixel to be processed in image data generated by a pixel array in which blocks, each including a plurality of pixels adjacent to each other, are arranged in a predetermined pattern. A prediction value for the pixel to be processed is then derived by applying the prediction method set, and a prediction residual obtained by subtracting the prediction value from each of pixel values in the image data is encoded.