Value correction is an automated data-cleaning feature in Breeze designed to identify and repair bad or out-of-bounds pixel values. When enabled it prevents saturated pixels, dead sensor elements and non-physical values from distorting spectral visualizations and predictions. Settings for the default value can be enabled/disabled in the General tab of Workspace Settings. It can be enabled/disabled for individual projects in the Project options.
How it works
The value correction operates through a two-stage process:
-
Static bad pixel masking.
Before scanning begins, Breeze analyzes the captured data-cubes for the Dark and White References:-
Statistical Variance Analysis:
Breeze calculates the mean frame and standard deviation across all sensor pixels. -
Standard Error & Outlier Thresholding:
Pixels exhibiting excessive noise variance, standard error or intensity deviations above strict statistical thresholds are flagged as defective dead/bad pixels. -
Static Mask Creation:
Pre-identified bad pixels are compiled into a static mask frame and registered to ba automatically corrected on every incoming scan frame.
-
-
Dynamic out-of-bounds correction.
During live processing, calculated spectral values are evaluated in real time: -
-
Range Validation:
In Spectral imaging | Reflectance mode, calculated values are checked against upper saturation limits. When calculating Spectral imaging | Absorbance, values below zero are also flagged to avoid logarithmic errors. -
Value Zeroing:
Any out-of-bounds pixel values is reset to zero and added to the block’s correction index for that frame. -
Neighboring Channel Interpolation:
For every flagged bad pixel, Breeze inspects adjacent spectral bands. If valid positive values are present, the bad pixel is replaced with the average of its immediate spectral neighbors.
-