Adaptive Intra-Prediction for Distorted Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-quality image compression is inefficient due to significant image distortion from lenses used in 360-degree cameras or drones, leading to increased residual values during intra-prediction, which reduces compression efficiency.
Innovation Solution
An electronic device with a lens unit, image sensor, and processor that divides images into blocks, selects blocks for intra-prediction based on distortion parameters, and uses direction modes with varying curvatures to minimize residual values, thereby improving compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a linear prediction method is used for intra-prediction of a spherically distorted image, then the processing complexity is low, but the residual value increases and compression efficiency decreases
Solution Approach 1:
The patent applies curved prediction lines with varying curvatures to match the spherical distortion characteristics of the image. Instead of using straight linear prediction lines, the invention introduces curved lines that bend according to the distortion parameter, allowing the prediction to follow the natural curvature of the distorted image regions and reduce residual values.
Solution Approach 2:
The patent dynamically adjusts the curvature of prediction lines based on the distortion parameter of different image regions. The curvature is not fixed but varies according to the local distortion characteristics, allowing the prediction method to adapt to the spherical distortion and maintain low complexity while improving compression efficiency.
2Manufacturing precision
If high-resolution and high-quality image data is transmitted or stored, then the image quality is improved, but the transmission cost and storage cost increase
Solution Approach 1:
The patent changes the prediction method parameter from linear to curved based on the distortion parameter. By adjusting the curvature parameter of the prediction lines to match the image distortion, the method achieves better prediction accuracy and lower residual values, thereby improving compression ratio while maintaining high image quality.
3Speed
If image compression is performed on distorted images using conventional methods, then the processing speed is fast, but the compression ratio is poor due to increased residual values
Solution Approach 1:
The patent introduces curved prediction lines that match the spherical distortion of the image, allowing faster compression with better ratios by reducing residual values through geometry-adaptive prediction.
Solution Approach 2:
The patent dynamically changes the curvature parameter of prediction lines based on local distortion characteristics, enabling efficient compression with improved ratios by adapting to image geometry without significant computational overhead.
Data Source
AI summary
Various embodiments provide an electronic device and a method for controlling an operation of the electronic device. An electronic device according to various embodiments of the present invention may comprise: a lens unit; an image sensor; a memory used for intra-prediction of an image captured using the image sensor, and storing direction modes differently configured according to characteristics of a partial area of the lens unit; and a processor, wherein the processor: acquires a raw image corresponding to an external object by using the image sensor; divides the raw image into multiple blocks; selects a block, on which intra-prediction is to be performed, from among the multiple blocks; determines a direction mode to be used for intra-prediction of the selected block on the basis of a distortion parameter of a partial area of the lens unit corresponding to the selected block; performs the intra-prediction on the selected block on the basis of the determined direction mode; and generates a compressed image corresponding to the raw image on the basis of a result of the intra-prediction. In addition, various embodiments are possible.