A fundamental image processing function is called blob analysis. It is the extraction of features from connected pixels that share the same logical state (blobs). MVTec’s software products contain a comprehensive collection of operators and tools that provide extreme flexibility and performance for the segmentation and analysis of many types of object features. MVTec’s software allows developers to define blobs of arbitrary shape (even if they are not connected) as Regions of Interest to which subsequent image processing operations are applied. This powerful method significantly reduces processing overhead and improves efficiency and execution speed compared to operating on regions defined by basic shapes like polygons or circles.
MVTec software performs blob analysis within milliseconds. Even when dealing with more complex images showing, for example, partially overlapping parts you want to measure. MVTec software provides many flexible tools for image segmentation like hysteresis, local, binary, and standard thresholding, along with over 20 additional segmentation operators. It also includes analysis functions for determining area, orientation, and about 50 more shape and gray value features.
A blob analysis mainly consists of three steps:
1. Acquire Image(s):
An image is acquired.
2. Segment Image(s):
Isolating the foreground pixels of interest from the image background using preprocessing tools and operations like thresholding and others. This is also called segmentation.
3. Extract Features:
Features like area (i.e., the number of pixels), center of gravity, or the orientation of a blob or blobs are calculated.
The figure on the right shows tissue particles in a liquid. These particles are bright and the liquid (background) is dark. By selecting bright pixels (thresholding) the particles can be detected easily.
In many applications the contrast of dark and bright pixels may not be as good, but the same results can be achieved with extra pre-processing or alternative methods for pixel selection / grouping.