| 摘要 |
A computer implemented method is disclosed for distinguishing between animals depicted in one or more images, based upon one or more taxonomic groups. The method comprises receiving image data depicting a respective animal, determining one or more spectral properties of at least some pixels of each image, and applying a k-means clustering algorithm on the determined spectral properties to allocate each image to one of a plurality of sets. The spectral analysis may include generating histograms and comparing the locations of peaks in respective spectral histogram data with reference data. The initial identification of animals may be carried out by applying an edge detection filter to a greyscale representation of the selected image.
|
