Object recognition

The input image is converted to grayscale and the contrast is improved if necessary. The objects are then segmented, i.e. the separation between object and background pixels. After the blob analysis, the specified measurement filters are applied so that only the desired objects (individual fibers and clusters) are displayed.

Cluster versa Single Fiber by Classification

The distinction between single fibers and fiber clusters is made on the basis of its shape. Fiber clusters are always wider than the single fiber and they have more than two endpoints. These factors can be used for the classification. Objects that are not fibers or fiber clusters can be excluded again, these are referred to as dust.

Cluster separation

The cluster images are binarized and the white pixels are displayed in Hough space. The individual fibers can now be calculated from the cluster image from the maximums occurring there. Each individual fiber is differentiated with a colored overlay. All recognized fibers can now be selected via a checklist and attached to the data table. Individual fibers that are not automatically recognized can be analyzed manually using the so-called Maxima click.

Hardware-Features:

System Configuration: CCD flatbed scanner with special lighting (I-Lumino-Box)
Sample supply: The fibers are transferred to a Petri dish and the Petri dish is placed on the scanning surface. The I-Lumino-Box moves along a rail and is placed exactly over the Pertri shell.
Preparation area: Petri dish - 90 mm inner diameter
Acquisition area: The scanning area is the entire Petri dish or a user-defined area.

Software-Features:

Resolution and smallest fiber thickness: 2400 dpi standard resolution - 10 µm smallest fiber thickness
Measurement range of the fiber length: 10 - approx. 40.000 µm
Border effects: none, as the complete Petri dish is scanned (edge ​​effects occur when the acquisition area is smaller than the preparation area)
Crossed straight fibers (clusters): automatic and manual analysis possible
Crossed curved fibers (long fiber clusters): automatically with the so-called ALF algorithm
Analysis duration: approx. 5 min. (without incineration and dilution - see preparation techniques)
Max. Number of fibers in an analysis: several thousand