After having successfully segmented and rotated the individual chromosomes from the input provided, also referred to as native image, the next step performed was to merge all the chromosome image files into a single image. This involved segmenting only the chromosomes around their boundaries and copying them onto a blank image file so that all the 46 chromosome objects appeared vertically aligned on a white background. Also, enhancement was performed which included contrast stretching. Also, a user interface was designed which kept track of the total number of detected objects. The application software provided functionalities for manually deleting and retaining the residual artifacts, undo-operation for mistaken deletions, and upside down flipping of chromosomes. Later on, chromosomes were grouped according to their length for initial karyotyping.
The next stage is to pair the individual chromosomes according to their unique band patterns. Many pattern matching algorithms for digital image processing are listed. However, for images with not very strong features, template based approach has proved to be favorable. The globally accepted notations of band patterns called ideograms are taken as reference images for extracting chromosome regions. For template matching, the projection vectors of the native chromosome images are computed by the average intensity profiles and band patterns are matched.
Automatic Karyotyping 