AI Filter Classification for Real-Time Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing technologies face challenges in achieving real-time processing while maintaining image quality, as non-learning-based methods are fast but inflexible, and learning-based methods are effective but complex, making them unsuitable for real-time applications like television and system-on-chip implementations.
Innovation Solution
An electronic apparatus and method that perform learning-based image processing in real time by classifying image blocks into patterns using a processor, applying filters learned through an artificial intelligence algorithm, and calculating gradient vectors to enhance image resolution and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If non-learning-based image processing methods are used, then processing speed is rapid, but flexible image processing depending on image characteristics is impossible
Solution Approach 1:
The patent implements dynamic filter selection based on image characteristics. The processor determines image patterns (edge, texture, flat area) and dynamically selects appropriate filters from multiple learned filters, allowing the system to adapt to different image types while maintaining real-time processing capability. This resolves the contradiction by making the processing method dynamic rather than static.
Solution Approach 2:
The patent segments the image processing task by dividing images into different patterns (edge, texture, flat area) and applying specialized filters to each segment. This segmentation allows flexible processing of different image characteristics while maintaining overall processing speed through parallel or sequential processing of segments.
2Adaptability or versatility
If learning-based image processing methods are used, then flexible image processing is possible, but real-time processing is difficult
Solution Approach 1:
The patent performs preliminary action by pre-learning and storing multiple filters corresponding to different image patterns before actual processing. During real-time processing, the system only needs to classify the image pattern and retrieve the pre-learned filter, avoiding the need for complex real-time learning calculations and enabling real-time processing.
Solution Approach 2:
The patent segments the learning process into offline pre-learning phase and online execution phase. The complex learning is done beforehand to create a lookup table of filters, while the online phase only involves simple pattern classification and filter application, achieving real-time processing capability.
3Device complexity
If conventional image processing methods are used, then processing is simple, but image quality improvement is limited
Solution Approach 1:
The patent changes the parameters of the processing system by introducing multiple learned filters with different characteristics (sharpness, smoothness, edge enhancement) and selecting the optimal filter based on image pattern. This parameter change enables significant image quality improvement while keeping the processing logic relatively simple through pattern-based selection.
Data Source
AI summary
An electronic apparatus is provided. The electronic apparatus includes: a storage configured to store a plurality of filters each corresponding to a plurality of image patterns; and a processor configured to classify an image block including a target pixel and a plurality of surrounding pixels into one of the plurality of image patterns based on a relationship between pixels within the image block and to obtain a final image block in which the target pixel is image-processed by applying at least one filter corresponding to the classified image pattern from among the plurality of filters to the image block, wherein the plurality of filters are obtained by learning, through an artificial intelligence algorithm, a relationship between a plurality of first sample image blocks and a plurality of second sample image blocks corresponding to the plurality of first sample image blocks based on each of the plurality of image patterns.