Network Camera Dynamic Range via Segmented Region Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance systems face limitations in achieving high intra-frame dynamic range in images, particularly in highly illuminated areas, and struggle to improve signal-to-noise ratio in dark areas without degrading frame rates.
Innovation Solution
High-resolution network cameras with on-board frame buffers and image processing units utilize multi-frame processing by identifying oversaturated image windows, performing auto exposure iterations, and reading out specific sub-frames with different optical integration times to create composite images with enhanced dynamic range, while improving signal-to-noise ratio through localized averaging and binning in dark areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If multiple complete frames are read out with different integration times to create wide dynamic range composite image, then intra-frame dynamic range is improved, but frame rate degrades and delays increase
Solution Approach 1:
The patent divides the image into multiple regions of interest (windows) and processes only those specific areas that require dynamic range enhancement. Instead of reading out entire frames multiple times, the system identifies oversaturated regions and performs multi-frame processing only on those segmented areas, thereby maintaining overall frame rate while improving dynamic range where needed.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image based on their specific characteristics. Oversaturated regions undergo multi-frame composite processing, while well-exposed regions are processed normally. This localized approach ensures that dynamic range enhancement is applied only where necessary, avoiding the performance penalty of processing the entire image uniformly.
2Illumination intensity
If multiple complete frames are read out with different integration times, then dynamic range enhancement is achieved, but processing time increases
Solution Approach 1:
The patent segments the image processing task by identifying and isolating only the oversaturated regions that require dynamic range enhancement. By processing only these specific windows rather than entire frames, the processing time is significantly reduced while still achieving the desired dynamic range improvement in the critical areas.
Solution Approach 2:
The patent performs processing on only the necessary portions of the image (partial action) rather than the entire frame. By applying multi-frame composite processing only to windows containing oversaturated pixels and leaving other areas unchanged, the system achieves adequate dynamic range enhancement with minimal processing time overhead.
3Measurement precision
If full frame readout is performed for dark areas to improve signal-to-noise ratio, then signal-to-noise ratio improves, but frame rate degrades
Solution Approach 1:
The patent identifies and segments dark regions from the rest of the image, then applies localized averaging (binning) only to those specific areas. This segmented approach allows signal-to-noise ratio improvement in dark regions without requiring full frame readout, thereby maintaining frame rate while achieving the desired noise reduction in critical areas.
Solution Approach 2:
The patent applies different processing techniques to different regions based on their characteristics. Dark regions receive localized averaging to improve signal-to-noise ratio, while other regions are processed normally. This local quality approach ensures that noise reduction is applied only where needed, avoiding the frame rate degradation that would result from processing the entire image.
Data Source
AI summary
A network camera comprising an image sensor, an image processing unit, a buffer memory and a network interface. The image processing unit is configured to divide the entire image readout by the image sensor into a plurality of overexposed regions and a plurality of non-overexposed regions, select a subset of the overexposed regions containing most oversaturated pixels, control said image sensor to readout only said selected subset of the overexposed regions with adjusted optical integration time, apply multiplicative scaling to readout pixels of said selected subset of the overexposed regions to generate scaled pixels, and replace pixels in the entire image corresponding to said selected subset of the overexposed regions with respective said scaled pixels to increase dynamic range of the entire image.