Image Encoding Residual Block Subdivision and 1D DCT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image encoding methods, such as those in H.264 and JPEG standards, face inefficiencies in prediction and compression due to inadequate utilization of horizontal or vertical correlations in residual data, leading to suboptimal compression efficiency and image quality, especially for pixels distant from neighboring blocks.
Innovation Solution
The method involves dividing residual blocks into sub-residual blocks, generating prediction sub-residual blocks using neighboring residues, calculating differences, and performing one-dimensional discrete cosine transformation (DCT) on these differences in lines, thereby improving prediction and compression efficiency by aligning with image characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If two-dimensional DCT is performed on residual data in 4×4 or 8×8 blocks, then compression is achieved, but horizontal or vertical correlations in the residual data are not efficiently utilized
Solution Approach 1:
The patent divides the residual block into multiple sub-residual blocks (e.g., 4 sub-blocks of 2×2 pixels each for an 8×8 residual block). This segmentation allows the transformation to be performed on smaller units that better preserve the horizontal or vertical correlation structure, thereby improving compression efficiency while adapting to the specific correlation characteristics of the image data.
2Ease of manufacture
If prediction is performed using only top and left neighboring pixels, then prediction is simple, but pixels far from neighboring blocks have large differences between prediction values and original pixel values
Solution Approach 1:
The patent extends prediction from using only top and left neighboring pixels to utilizing neighboring pixels in multiple directions (top, left, top-right, bottom-left, etc.). This dimensional expansion of the prediction neighborhood allows pixels far from the original reference pixels to be predicted more accurately by leveraging correlations from multiple spatial directions, thereby improving prediction accuracy without significantly increasing complexity.
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 enhances compression efficiency by reducing the magnitude of transformation coefficients and improving prediction accuracy, leading to better image quality and reduced bit rates, particularly when horizontal or vertical correlations are present in residual blocks.
Implementation Method 1
performing a one-dimensional discrete cosine transformation (DCT) on these differences in lines
Data Source
AI summary
Provided are a method and apparatus for image encoding which improves encoding efficiency in accordance with image characteristics by performing prediction in lines and performing a one-dimensional transformation in lines on an input image, and a method and apparatus for image decoding. Encoding efficiency of an image may be improved by generating a prediction sub residual block using neighboring residues and performing a one-dimensional discrete cosine transformation (DCT) on a difference residual block which is a difference between an original sub residual block and the prediction sub residual block.