Inverse Transform Cropping for Sparse Frequency Coefficient Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inverse transform processing in image and video coding requires significant computational resources and efficiency, particularly due to the need to calculate non-zero and zero frequency coefficients, leading to resource waste and inefficiency.
Innovation Solution
Determine a non-zero block in the frequency coefficient matrix and perform inverse transform calculations only on this block, using a cropped transform and transposed submatrix, thereby reducing unnecessary calculations on zero coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inverse transform processing is performed on the entire frequency coefficient matrix, then the complete image data can be reconstructed, but the calculation amount increases significantly and processing efficiency decreases
Solution Approach 1:
The patent extracts only the non-zero frequency coefficients from the frequency coefficient matrix and performs inverse transform processing solely on this extracted submatrix. This is achieved by first identifying the positions of non-zero coefficients, then constructing a reduced transform matrix corresponding only to these non-zero elements, thereby eliminating unnecessary calculations on zero coefficients while maintaining complete image data reconstruction capability
Solution Approach 2:
The patent segments the frequency coefficient matrix into non-zero blocks and zero regions. By dividing the matrix into meaningful segments based on the positions of non-zero coefficients, the system can process only the relevant segments (non-zero blocks) while ignoring the zero regions, thus reducing overall computational complexity without affecting reconstruction quality
2Reliability
If inverse transform processing is performed on all frequency coefficients including zero coefficients, then mathematical completeness is maintained, but computational resources are wasted
Solution Approach 1:
The patent extracts and processes only the non-zero frequency coefficients, eliminating the need to perform calculations on zero coefficients. This extraction approach maintains mathematical completeness for the relevant data while significantly reducing computational resource consumption, as zero coefficients contribute nothing to the final reconstructed image data
Solution Approach 2:
Instead of performing complete inverse transform on the entire matrix (excessive action), the patent applies partial inverse transform only to the necessary non-zero coefficient regions. This partial action is sufficient to achieve the desired reconstruction without the waste of computing resources on zero coefficients
Data Source
AI summary
The present disclosure provides an image processing method and apparatus, a device, and a medium, and relates to the field of image and video processing technologies. An implementation solution includes: performing frequency domain transform on a target image and obtaining a frequency coefficient matrix of the target image, the frequency coefficient matrix including at least one non-zero frequency coefficient; determining a non-zero block in the frequency coefficient matrix based on a position of the at least one non-zero frequency coefficient; determining a corresponding transform submatrix and a corresponding transposed submatrix of the non-zero block; and determining an inverse transform result of the frequency coefficient matrix based on the transform submatrix, the non-zero block, and the transposed submatrix.