From image to count

Scientific Counter is designed for analysis tasks where digital images must be evaluated systematically. The key settings are optimized in a test analysis and then applied to the entire image.

1. Load an image. Open a representative image from the test series.
2. Set the color selection. For dark objects, the image can be inverted. Select the red, green, blue, or grayscale channel in which the objects are most visible.
3. Define the analysis area. Circular inclusion areas, edge exclusion, and test areas help hide petri-dish edges, holders, or irrelevant image regions.
4. Prepare the image. Median filters remove brightness gradients, mean filters reduce small disturbances, and the threshold separates object regions from the background.
5. Select objects. Size, brightness, and shape factor can be specified as minimum and maximum values.
6. Separate overlapping objects. Automatic region separation can split connected object groups into sub-objects. Criteria such as convexity, size, and shape factor control which objects are separated.
7. Review the test analysis. Step-by-step test analysis shows intermediate results so parameters can be improved iteratively.
8. Set scale and output. A known scale enables size values in units such as µm or mm. Result files, histograms, and report output are defined in the options.

Quality starts during image acquisition

The software can only detect features that are sufficiently visible in the image. For stable results, lighting, camera settings, and camera distance should remain as constant as possible within an image series. For very small structures, lossless image formats are often better than JPEG because compression artifacts can interfere with detection.

In which order are the analysis parameters applied?

The analysis settings are applied in a fixed order. This is important because some parameters prepare the image, others control object detection, and some are only used at the end to decide whether a detected object is included in the results.

1. Define the analysis area. First, Scientific Counter determines which part of the image is evaluated. Inclusion circles, exclusion rectangles, edge exclusions, or test areas can hide irrelevant image regions such as petri-dish edges, holders, labels, or empty border areas.
2. Prepare the image. The image preparation settings are then applied. These include, for example, background filtering, absolute image difference, mean filtering, and median filtering. These steps improve the image before objects are detected.
3. Determine the object threshold. Next, the object threshold is calculated or a fixed object threshold is used. The threshold separates object regions from the background.
4. Optionally correct artifacts. If artifact correction is enabled, very unusual regions can already be excluded before the final object evaluation. This step can use the artifact threshold and the form factor limit.
5. Detect objects. Connected image regions are now marked as objects. If enabled, inclusions or holes inside objects can be removed at this stage.
6. Optionally separate connected objects. If area separation is enabled, Scientific Counter checks whether connected object regions should be split into separate objects. The exclusion criteria are considered in sequence, for example convexity, size, and form factor.
7. Calculate object parameters. For each detected object, the software calculates values such as area, brightness, perimeter, form factor, Feret diameters, circularity, and other shape parameters.
8. Apply the final object selection criteria. At the end, the detected objects are checked against the selected minimum and maximum values. The final evaluation is performed in this order: brightness, size, edge object, form factor.

In practice, this means that an object must first be detected and measured before limits for brightness, size, or form factor can be applied. Filters and thresholds influence what is detected as an object. The final object selection criteria then decide whether this object is accepted or excluded from the result.

If an object matches several exclusion criteria at the same time, a criterion checked later can overwrite an earlier evaluation. Therefore, the displayed exclusion reason is not always the first property that was outside the selected limits.