OpenCV

Most important features:

  • computer vision and machine learning software library
  • more than 2500 optimized algorithms
  • detect, recognize faces, identify objects, classify human actions in videos
  • track camera movements, track moving objects, extract 3d models of objects
  • produce 3d point clouds

For more information visit: OpenCV Webpage

Itom offers the full power of python 3 in combination with OpenCV 3. All OpenCV functions can be easily used and the results are visualized and further processed (e.g. output to Matlab) in python.

OpenCV is the standard open source image processing library. It is used in many applications for low-level image processing tasks up to dedicated computer vision applications. It can be used in C++, Java as well as python even for commercial projects. The methods are implemented in a very efficient manner and can be used with all sorts of images (color, monochrome and different depths).

Plenty of books for learning OpenCV are available and for a fast “Getting Started” with image processing using itom and OpenCV you might just have a look at Digitale Bildverarbeitung .

OpenCV offers all standard image processing methods like:

  • Linear filtering
  • Global transforms (especially Fourier)
  • Local operations
  • Morphological image processing
  • Camera calibration and geometrical transforms
  • Non-linear filters
  • Feature detection methods
  • etc.

… enough to let you really implement sophisticated image processing applications with a minimum of overhead.

Examples

The image shows noise reduction while keeping small details (in this case the peak at right side of the intensity distribution) using a bilateral filter.

bilateral filter

Something more sophisticated: Haar Cascades for feature detection (face detection):

face detection

And a local histogram equalization in order to make local features visible (see the edge of the cup and the wood texture).

histogram equal

The following screenshot shows the application of the so-called Hough transform used to find (in this case: long) lines in a very noisy image.

Hough Transform

Often, edge detection and noise reduction are first applied in order to perform further processing like the Hough transform. The following image shows two often used versions for edge enhancement or edge detection: Laplacian filtering and the famous Canny filter.

edge detection

Books