RGB Filter

For example, if Red is chosen:

R = ((R-B)+(R-G))

G = 0

B = 0

R is then normalized with respect to the maximum red value.

Based on the above formula it can be seen that white pixels result in a zero value whereas pure primary colors (R=255, G=0, B=0) R doubles its value. Thus function does a better job than RGB Channel in filtering for a particular color as white light is removed.

Due to normalization really dark pixels can be elevated in intensity and generate too much noise in the resulting image. The Min Pixel Value allows you to specify a minimum value below which pixels are considered to be black and will be ignored when calculating the image results. Default value is 40 (0-255).

You can use this filter to focus the image towards certain colors even with diminished lighting conditions.

Interface

Instructions

1. Colors - Chose the desired color Red, Green, Blue, etc. by selecting the appropriate checkbox. Note that "Colors" refers to those pixels that have strong colors in the image regardless of what color the pixel is (color saturation) while "Gray" refers to how close to a gray color pixels are. These two selections can be used to diminish colors and pronounce gray areas ("Gray") or diminish white areas and pronounce colors ("Colors").
2. Threshold - Select the min intensity pixel value that specified which values should be removed that are below a certain threshold. This helps remove pixels that are dark and do not contain enough color information. This is typically seen when detecting blue in dark areas.
3. Threshold - Select the hue threshold. This removes colors that are not 'blue' enough or not 'red' enough, etc.
4. Threshold - If needed select a hysteresis level. The hysteresis level will allow blobs that have at least one pixel above the hue threshold to grow into the surrounding area as long as the hue level is above the threshold minus the hysteresis level. Thus if the hue threshold is set to 80 and the hysteresis is set to 20 then only blobs that have a pixel above 80 will be preserved and any that do will include all pixels with hue above 60. This is useful to grow/connect a blob below the hue threshold while still preserving the filtering effects of the hue level.
4. Results - Select how the results should be represented
RGB Scale - resulting values are scaled RGB values from 0 to 255 that depend on how close the original color is to one in the Colors list
RGB Mask - resulting value is the color that is most similar to one in the Color list
Grayscale - resulting value is how close the color is to a color in the Colors list but represented in grayscale values
White Mask - resulting match value is white, non-match is black
Black Mask - resulting match value is black, non-match is white

Example

Source Image	Green Filter with 10, 30 threshold

See Also

Color Filter
RGB Channel
Threshold
Auto Threshold