*Local Threshold

Segmentation > *Local Threshold

Requires B/W Companion Image

Selects pixels if they meet the specified criteria relative to each feature in the Companion Image. “Auto Threshold” mode calculates the Otsu-based optimum threshold for each feature. “Percentage Of” and “Departure From” asses each pixel relative to its feature’s mean or median pixel value. “Upper Percentile and “Lower Percentile” select the pixels within the specified percentile of their feature’s mean or median pixel value.

1. Mode

Selects which mode to use for local thresholding

• Auto Threshold: Selects pixels if they are below the auto-threshold value for each region determined using Otsu’s method [1]
• Percentage Of: Selects pixels if they are below a certain percentage of their region’s critical statistic
• Difference Below: Selects pixels if they are below a certain difference from their region’s critical statistic
• Lower Percentile: Selects pixels if they are with in the specified lower percentile of their region’s grayscale pixel values
• Upper Percentile: Selects pixels if they are with in the specified upper percentile of their region’s grayscale pixel values

2. Statistic

Selects which statistic to consider for the current mode

• Mean (for Percentage Of and Difference Below): Compares pixels to their region’s mean grayscale intensity
• Median (for Percentage Of and Difference Below): Compares pixels to their region’s median grayscale intensity

3. Value

• Adjustment (for Auto Threshold): Offsets each region’s auto-determined threshold value by this amount
• Percentage (for Percentage Of): Specifies the critical percentage of the region’s mean or median each pixel must be below
• Difference (for Difference Below): Specifies the critical difference from the region’s mean or median each pixel must be below
• Percentile (for Lower Percentile): Specifies the lower percentile of the region’s grayscale values that pixels must be within
• Percentile (for Upper Percentile): Specifies the upper percentile of the region’s grayscale values that pixels must be within

References

[1] Otsu, N., “A Threshold Selection Method from Gray-Level Histograms,” IEEE Transactions on Systems, Man, and Cybernetics, Vol. 9, No. 1, 1979, pp. 62-66.