Our FoveaPro software provides scientists a wide array of tools for image processing and analysis including advanced refinement, measurement, and counting
a process of image refinement and measurement:
Below is a set of image series where we start with an original image and go through a number of processing steps to finally isolate the measurements that are important for a given process.
Original rice image
Using a color lookup table to show the contrast better,
We can see that the image has uneven illumination. Some of the rice grains are brighter than the others and we need to fix this before we can isolate them.
This can be corrected with any number of methods. We'll use AutoLevel for this example:
With this as the newly leveled image:
Leveling helps because now we can threshold the image more consistently to isolate all of the rice, instead of just the ones in the middle or on the top and bottom.
This shows a methods of Automatic thresholding that will separate the rice grains from the background into a simple image that just contains the features we wish to measure.
Here are the rice grains labeled. There are 53 of them that do not touch the edge of the image. (We don't measure edge-touching features for a variety of statistical reasons.)
This is an image of Latex spheres that we need to compute the distribution of diameters.
First, we need to threshold the image to isolate the spheres.
Unfortunately, the spheres are touching so we need to use a technique called “Watershed Segmentation” to cut them apart.
Now, we can measure the side distribution of the circumscribed radii (if we are to assume that they are spheres, the largest radii is mostly likely to be correct).
The following image has those radii drawn on top to illustrate the process.
Check out our other examples and a thorough tutorial on FoveaPro.
For more information: