All the algorithms used by Astra are available in IPSDK.
Astra allows to easily configure image processings giving a large choice of approaches. The optimized implementation ensures optimal execution speeds.
Image processing features
Astra offers the possibility to pre-process your image smoothing it. For this, the available filters are :
separable bilateral filter,
anisotropic diffusion filter.
it is possible to configure these filters in order to fit to the application needs.
Once the image is pre-processed, it is possible to display the original image or its smooth version with the corresponding tab.
The first processing step is the image binarization. Astra proposes to select features in the image according to their intensities (bright features on dark background or dark foreground on bright background).
It is also possible to manually define the binarization threshold or to apply an Otsu's automatic thresholding.
Finally, a preview button allows to display the binarization result in order to validate its result before the rest of the process.
The next step consists on segmenting the binarized image. Astra proposes to use (or not use) a filter to separate features, to avoid the process of features positionned at the border of the image and to remove the holes in each object.
Once the parameters are defined, a button allows to display the segmentation result in order to validate this step.
Astra makes available a large number of features, provided by IPSDK, on the previously segmented features.
The available measures can be geometric measures (such as the area, the perimeter or the bounding box) or intensity-based ones (such as the minimum, maximum or mean intensities).
Astra allows to calibrate the measures. To do this, it is just necessary to specify the pixel size and the unity (meter, millimeters, etc.) in the corresponding fields.