Updating Fixed Pattern Noise Matrix for Video Stream Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods are inadequate for effectively identifying and reducing fixed pattern noise in images, as it remains constant relative to the image sensor and varies in signal strength over time, making calibration and averaging ineffective.
Innovation Solution
A method for updating a fixed pattern noise matrix by calculating differences between pixel values in an electronic-image stabilized video stream, identifying candidate positions for noise, and adjusting these positions based on stabilization data to improve noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If calibration is performed to remove fixed pattern noise, then fixed pattern noise reduction is improved, but the method is not suitable because signal strength varies over time
Solution Approach 1:
The patent implements dynamic updating of the fixed pattern noise matrix by continuously calculating differential matrices from video stream images and identifying candidate positions where pixel value differences deviate from predetermined values. This dynamic approach allows the system to track and update fixed pattern noise characteristics over time, making the noise reduction adaptive to signal strength variations while maintaining calibration effectiveness.
2Measurement precision
If averaging multiple images is performed to reduce noise, then random noise reduction is improved, but fixed pattern noise reduction is not suitable because noise location is fixed
Solution Approach 1:
The patent extracts fixed pattern noise from images by calculating differential matrices between target images and reference images, identifying candidate positions where pixel value differences exceed predetermined thresholds. By separating the fixed pattern noise identification process from the image content, the system can selectively remove fixed pattern noise while preserving random noise reduction benefits through subsequent processing.
3Measurement precision
If fixed pattern noise matrix is updated using differential matrices, then identification accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies a predetermined threshold value to identify candidate positions in differential matrices, focusing computational resources only on positions where pixel value differences exceed this threshold. This partial action approach processes only the necessary portions of the differential matrices rather than analyzing all pixel positions, thereby improving identification accuracy while limiting the increase in computational complexity to only the regions of interest.
Data Source
AI summary
A method for updating a fixed pattern noise matrix comprises: calculating a first difference between a target and first different images in a video stream to obtain a first differential matrix; calculating a second difference between the target and second different images in the video stream to obtain a second differential matrix; identifying a set of candidate positions for fixed pattern noise by: locating first and second sets of positions in the first differential matrix at which a difference deviates from predetermined values, finding a set of overlapping positions between the first and second sets of positions, and adjusting the set of overlapping positions. The adjusted set of overlapping positions is used for fixed pattern noise. Furthermore, each position in the set of candidate positions is updated, wherein the updated fixed pattern noise value at each position is based on a value at a corresponding non-adjusted position in the differential matrix.