IDOL Image Server extracts information from images, returning the results as XML documents, which you can index into IDOL Server. You can also use this information to tag images, which makes it easier to retrieve relevant images.
Image Server supports the standard image formats TIFF, JPEG, PNG, GIF, BMP, PBM, PGM, PPM. It also supports PDF, using IDOL KeyView.
Image Server can detect and extract text from images. The form of the input can be:
documents (computer generated, printed, or scanned)
frames from subtitled television footage, including news channels with banners
photographs of natural scenes, for example of street signs
photographs of documents, for example, from a smartphone
Image Server can detect human faces in images. It can compare detected faces against user-created databases to recognize known faces.
You can also perform demographics analysis on faces, to determine gender, age, race, and facial expression.
Image Server can find the location and color of a person's clothing in an image, if you supply the location of the face in the image.
Image Server can detect most types of barcode in documents, and decode them. It also supports analysis of photographs of documents (for example, from a smartphone).
Image Server can detect and recognize objects in documents, such as printed logos, product packaging, and adverts. It compares the results to user-supplied examples.
Image Server also supports logo analysis for photos and TV frames.
Image Server can compare images of redacted documents to the original, un-redacted text, and mark the differences.
Image Server can perform simple color-related analysis of images, to find:
the dominant color of an image.
the main central color of an image.
the proportion of skin color in an image (this ability provides a very limited detection of possible adult content).
a text hash of an image, which IDOL Server can use to find similar images.
|