Possible problems with Grain and Pattern counting

• No patterns detected Threshold is too high, mismatch weight is too negative, or match weight is too low; or, contrast is too low in image.
• Too many patterns detected Threshold is too low, mismatch weight is not negative enough, or match weight is too high. The optimum threshold is different for each type of image. The ideal threshold can be estimated visually by watching the numbers in the information box, and determining the signal at the point at which non-patterns begin to be detected. Interrupt the counting at this point and set the threshold to the value of the signal shown in the information box.
• Same pattern counted more than once; or, algorithm gets stuck on one pattern (coordinates don't change). This is a bug; please report it.
• Processing is too slow. Convert image to grayscale. This will increase speed at least 2.8 fold. Converting image to 8 bits will also help.
• Message box says ``0 Patterns found'', even though it found patterns on the same image earlier. You may have accidentally clicked on the background or menu bar, deselecting the image. Click on image and repeat.
• Message box says ``Neuron count mismatch''. This message occurs when you click on a different image while the algorithm is counting. The buffers are incorrect for the new image, so the algorithm aborts.
• Small red box indicating location of pattern is off-center. This occurs when defining the pattern, if the region selected for the pattern is not centered on the pattern (i.e., too much background is included in the pattern on one side).
• Large areas with shading similar to the pattern are identified as numerous instances of the pattern. This is caused by selecting an insufficient amount of background around the prototype pattern; or the mismatch weight is too low.

Examples

Size distribution using quick segmentation

Here is an example of an astronomical image after analysis by the quick segmentation algorithm. The threshold was set to 0.8 so that only the larger stars are counted. Each star that was counted is automatically labeled with its relative size. The size of a given object depends on the threshold and will be larger if the threshold is smaller, because more pixels are present in the distance matrix which is used to find grains. The grain counting dialog box is shown at right.

$\parbox{4in}{ \baselineskip 11pt The size distribution graph (not ... ...d verify that the results obtained are consistent with their expectations. }$

Recommended Threshold and Weight values for neural network method

Type of pattern	Threshold	Match Wt.	Mismatch Wt.
Grains (monochrome)	0.50	1.0	-0.05
Grains (color)	0.45	1.0	-0.10
Grains (enhanced)	0.55	1.0	-0.15
Faces	0.60	1.0	-0.15
Faces (color)	0.95	1.0	-1.00
Patterns	0.90	1.0	-0.50

Note: These values should be considered as starting points only. Different images will need different thresholds and mismatch weights.

Example of neural network grain counting:

Below is an example of neural network grain-counting of a moderately-difficult image illustrating some of the steps that may be necessary to obtain the most accurate counts. Panel A shows the original image, which contains several grains that are out of focus or clumped together, as well as some grains that are sharply focused, and a darker, out-of-focus blue cell in the center, which could be difficult to distinguish from the two faint grains on top of it. Under some conditions, the cell could even be misidentified as a large clump of grains. The image was converted to grayscale (B) and the grains were enhanced (by clicking on ``Enhance grains''), producing C, which is more easily analyzed. Using the default threshold value of 0.5 and Match and Mismatch weights of 1 and -0.05, however, caused a number of extraneous points to be counted as grains (D). This occurred because the filtering process also increased the noise slightly. Changing the mismatch weight to -0.15 solved the problem, and the program counted the number of grains as 64, the correct number (E). (Note: This figure may not display clearly in xdvi).