Dark Reference Pixel Noise Estimation Using Absolute Differences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating noise levels in image sensors are computationally expensive, sensitive to defective pixels, and less robust to manufacturing variations, requiring significant computation power and memory resources.
Innovation Solution
A method that estimates noise levels by sequentially computing and accumulating absolute differences between dark signal levels of adjacent pixels in image sensors, reducing the need for memory and computation power, and focusing on real measurements to minimize the influence of defective pixels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional standard deviation computation is used to estimate noise level, then measurement precision is improved, but device complexity and computation power requirements increase significantly
Solution Approach 1:
The patent changes the mathematical parameter from standard deviation (requiring mean calculation and squared deviations) to mean absolute difference (requiring only absolute difference calculation). This parameter transformation maintains noise estimation capability while dramatically reducing computational complexity and memory requirements
Solution Approach 2:
The patent extracts only the essential information needed for noise estimation by using absolute differences between adjacent pixels, discarding the computationally intensive mean calculation and squared deviation steps. This extraction approach retains measurement precision while eliminating unnecessary computational steps
2Measurement precision
If standard deviation computation with multiple passes is used, then measurement precision is improved, but loss of time increases due to processing lag
Solution Approach 1:
The patent enables continuous single-pass computation of noise levels by calculating absolute differences between adjacent pixels as data flows through, eliminating the time delay inherent in two-pass methods. This continuous action maintains precision while reducing processing time and latency
Solution Approach 2:
The patent performs the computation in a single forward pass through the pixel data, calculating absolute differences immediately as each pixel is processed. This preliminary action eliminates the need for a second computation pass, reducing processing time while maintaining measurement accuracy
3Measurement precision
If standard deviation computation is used, then measurement precision is improved, but reliability decreases due to sensitivity to defective pixels
Solution Approach 1:
The patent converts the potential harm of defective pixels (outliers) into a benefit by using absolute differences instead of squared deviations. The absolute difference metric is less sensitive to extreme values, so defective pixels have reduced impact on the noise estimation, improving reliability while maintaining precision
Solution Approach 2:
The patent changes the mathematical parameter from standard deviation (squared deviations) to mean absolute difference. This parameter transformation reduces the influence of outliers and defective pixels on the noise estimation, making the measurement more reliable while preserving accuracy for normal pixels
4Device complexity
If noise model with determined parameters is used, then device complexity is reduced, but reliability decreases due to sensitivity to manufacturing variations
Solution Approach 1:
The patent enables the image sensor to self-measure its own noise characteristics by computing absolute differences between adjacent dark pixels. This self-service approach eliminates dependence on pre-calibrated parameters and temperature sensors, improving reliability across manufacturing variations while keeping device complexity low
Solution Approach 2:
The patent implements a feedback mechanism where the actual dark pixel values from the image sensor are used to compute the noise level. This real feedback from the sensor itself adapts to manufacturing variations and environmental conditions, improving reliability compared to fixed noise models while maintaining computational simplicity
Data Source
AI summary
A system has an array of pixels including a plurality of active pixels and a plurality of dark reference pixels and processing circuitry coupled to the array of pixels. The processing circuitry sequentially computes, for each of a plurality of pairs of sets of dark reference pixels of the plurality of dark reference pixels, absolute differences in dark signal levels of the pair of sets of dark reference pixels. The absolute differences in dark signal levels are accumulated and a noise level of the dark reference pixels of the array of pixels is estimated based on the accumulated absolute differences. The system may be employed in, for example, a back-up camera of an automobile or a mobile phone.