Image Matching

LTU's image matching technology automatically and very efficiently finds matching images within a database of images. The technology can find images that are exactly the same, as well as images that have been edited in any way and photos taken with a mobile device.

LTU technologies' image matching technology uses distinct local image features to compute a unique image signature (also called image DNA). These signatures are used to identify matches between a query image and the reference database.

Image matching is highly configurable and thus applicable to a vast number of different applications that require image matching. The technology can be configured to be optimized for different applications with options such as rotation invariance, thresholds, text removal and flip invariance.

Cropped images - One image can be a "cropped" version of the other, with additional editions like color changing, reencoding, etc.. Composite image - One image can be made of several other images. The matching technology will detect all original images it was made of. Photometric and statistic noise -  hue, contrast, luminance, palette changes, posterization, resizing and reencoding Geometric transformations - resizing, length-width ratio change, flipping and rotation Photometric transformations - grayscale, contrast, color, hue, palette changes and many more... Filtering effects - noise, smoothing, paintbrush, median, and many more... Annotations - captioning and drawing on an image.