Pixel Clustering for Robust Image Source Combination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image and video processing systems face limitations in combining outputs from different estimation methods, particularly in rejecting outliers and choosing appropriate weighting schemes, leading to errors and requiring extensive tuning.
Innovation Solution
A technique that computes distance measures between pixel values, forms clusters, determines scores for each cluster, and selects one cluster to derive an output pixel value, using iterative merging and rule-based selection methods to enhance robustness and reduce resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a median of estimated values is used to combine multiple estimators, then simplicity and rejection of outliers are improved, but accuracy deteriorates when the correct value is an outlier
Solution Approach 1:
The patent segments the combining process into distinct phases: first forming clusters of similar pixel values, then evaluating each cluster separately using multiple estimation methods, and finally selecting the best cluster. This segmentation allows the system to handle outlier cases by isolating them in separate clusters rather than having them contaminate a simple median calculation.
Solution Approach 2:
The patent changes the parameter space by transforming individual pixel values into clusters of pixel values with associated scores. Instead of directly combining scalar values, the system creates a higher-dimensional representation where each cluster has characteristics (score, size, composition) that can be evaluated independently, allowing the correct outlier value to be identified through its cluster properties.
2Reliability
If a weighted combination of estimated values is used, then accuracy can be improved, but difficulty in choosing appropriate weighting scheme increases
Solution Approach 1:
The patent implements self-service by having the system automatically determine weights through cluster scoring rather than requiring manual weighting scheme selection. The scoring mechanism evaluates clusters based on their internal consistency and characteristics, automatically assigning appropriate weights without human intervention. This eliminates the complexity of choosing weighting schemes while maintaining accuracy.
3Reliability
If adaption between estimated values is used based on motion or edge characteristics, then accuracy can be improved, but measurement errors and algorithm parameter tuning requirements increase
Solution Approach 1:
The patent extracts the adaptation logic from complex motion and edge analysis into a simpler cluster-based framework. Instead of directly adapting weights based on motion characteristics, the system extracts pixel values into clusters and lets the cluster properties naturally reflect motion and edge information. This extraction reduces measurement errors and eliminates the need for extensive parameter tuning while preserving accuracy.
Data Source
AI summary
A technique is described for combining several image sources into a single output image or video sequence. For a given pixel of the output image, pixel values are received from the image sources, and a matrix of distance measures between the pixel values (e.g. based on their colors) is computed. Clusters of pixel values are formed using the distance measures, and a score determined for each. One of the clusters is selected according to the scores, and used to derive an output pixel value. In an example, the clusters are formed using an iterative process where the closest pairs of pixel values or clusters are merged to form new clusters up to a size threshold. Examples are described for scoring the clusters based on model-based weighting or cluster size. Examples are also described for a rule-based cluster selection system. A composite image generator implementing the technique is also described.