Sparse Image Quality Enhancement via Analytic Coefficient Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current learning-type image quality enhancement methods utilizing sparse expression face challenges in achieving real-time performance due to high processing delays and increased circuit costs, especially when enhancing image quality during image enlargement.
Innovation Solution
An image quality enhancing apparatus that calculates a sparse solution for coefficient matrices analytically, selecting fewer base vectors and minimizing the sum of squares of differences and coefficients, allowing for efficient image quality enhancement without iteration methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If learning-type image quality enhancement utilizing sparse expression is implemented, then image quality is improved, but processing delay increases and real-time performance deteriorates
Solution Approach 1:
The patent pre-calculates and stores correspondence relationships between low-quality and high-quality base vectors in a dictionary during an offline learning phase. This preliminary action allows the online enhancement process to simply lookup and combine pre-computed vectors, dramatically reducing processing delay while maintaining image quality enhancement capabilities
Solution Approach 2:
The patent segments the image enhancement process into distinct components: extracting feature quantities from the input image, selecting relevant base vectors from pre-computed dictionaries, and combining them with calculated coefficients. This segmentation allows each component to be optimized independently, with the dictionaries pre-computed offline and only the essential combination step requiring online processing
2Manufacturing precision
If learning-type image quality enhancement utilizing sparse expression is implemented, then image quality is improved, but circuit cost increases
Solution Approach 1:
The patent performs the computationally intensive learning process offline to pre-compute and store base vector correspondences in dictionaries. This shifts the complexity burden from the online circuit implementation to an offline computation phase, allowing the actual image enhancement circuit to use simpler operations like vector selection and linear combination
Solution Approach 2:
The patent transforms the complex sparse optimization problem into a simpler parameter selection problem by pre-computing optimal base vectors. The online system only needs to select from pre-determined options and combine them with calculated coefficients, significantly reducing the computational complexity and circuit requirements
3Measurement precision
If iteration method is used to calculate sparse solution, then accuracy is improved, but processing duration increases
Solution Approach 1:
The patent pre-calculates the optimal sparse solutions and stores them as correspondence relationships in dictionaries during an offline learning phase. This eliminates the need for iterative calculations during online operation, as the system directly applies pre-computed coefficients and base vectors, achieving both accuracy and speed
Solution Approach 2:
The patent creates simplified copies of the optimal solutions in the form of pre-computed dictionaries containing base vector correspondences. Instead of repeatedly solving the complex optimization problem, the system copies and applies the pre-stored solutions, dramatically reducing processing duration while maintaining accuracy
Data Source
AI summary
An image quality enhancing apparatus, an image display apparatus, an image quality enhancing method, and a computer readable storage medium which make a learning-type image quality enhancing method utilizing a sparse expression practical are provided. The image quality enhancing apparatus calculates, from the feature quantity of an image, coefficients of low-image-quality base vectors expressing a feature quantity with a linear sum and generates the image with the enhanced image quality by calculating a linear sum of high-image-quality base vectors using the calculated coefficient. When calculating the coefficient, T base vectors highly influential on the feature quantity are selected from among a plurality of base vectors and an analytic solution making L2 norm of a coefficient matrix α as small as possible is calculated. A sparse solution of the coefficients can be obtained without using the iteration method and a practical image quality enhancing apparatus can be realized.