Lenslet Image Predictive Coding Through Frequency Component Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression techniques are inefficient for lenslet images due to their macropixel structure, leading to high data volumes in light field displays, which are not adequately addressed by traditional codecs.
Innovation Solution
A system that determines reference pixel blocks and decomposes prediction blocks into components using machine learning models to predict pixel values, encoding the image data based on these components, reducing the data required for storage and transmission by using residuals and bitstreams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional natural image coding schemes are used to encode lenslet images, then the encoding process is simple, but the compression efficiency is inadequate due to the macropixel structure
Solution Approach 1:
The patent segments the prediction block into three distinct frequency components (DC component representing average pixel values, low frequency component representing low frequency pixel values, and high frequency component representing high frequency pixel values). This segmentation allows each component to be encoded separately, improving compression efficiency by exploiting the statistical properties of different frequency ranges in lenslet images with macropixel structures.
Solution Approach 2:
The patent transforms the encoding approach by changing the domain from spatial pixel values to frequency domain components. By applying frequency transformation and encoding separate components (DC, low frequency, high frequency) with different precision levels, the system achieves better compression efficiency while adapting to the unique macropixel structure of lenslet images.
2Loss of information
If all pixel values of prediction blocks are transmitted, then complete image data is preserved, but the data volume becomes extremely large for storage and communication
Solution Approach 1:
The patent extracts only the essential information from prediction blocks by representing them through three components: DC component (average pixel values), low frequency component, and high frequency component. This extraction approach captures the most important visual information while discarding redundant data, significantly reducing data volume while maintaining acceptable image quality for light field displays.
Solution Approach 2:
The patent discards redundant high-frequency details that are less perceptible to human vision, while preserving the essential low-frequency and DC components that carry the main image information. This selective discarding and recovery strategy enables significant data compression while maintaining the perceptual quality of lenslet images.
Data Source
AI summary
Systems, methods and apparatuses are described herein for accessing image data, generated at least in part using a device comprising a lenslet array, determining a plurality of reference pixel blocks of the image data, and determining a prediction block in a vicinity of the reference pixel blocks. Implementing any of the technique(s) described herein, the system or systems may determine, based on the plurality of reference pixel blocks, a first component representing average pixel values of the prediction block, a second component representing low frequency pixel values of the prediction block, and a third component representing high frequency pixel values of the prediction block. The system(s) may determine predicted pixel values of the prediction block based on the first component, the second component and the third component, and encode the image data based at least in part on the predicted pixel values of the prediction block.