Object recognition method based on use of multiple images of the same object the pixels of which are combined in an unstructured quantity from which image characteristics are extracted for object recognition

摘要

Method for object recognition in which a sequence of digital images of an object are recorded and each image converted into corresponding pixels with at least two images taken from different perspectives. The pixels of the images are converted into a common coordinate system and superimposed to form an unstructured whole pixel quantity which corresponds to the object. Lastly object characteristics are extracted from the pixel quantity with a function that samples the unstructured quantity so that an object can be recognized from the characteristics. An Independent claim is made for an image recognition system with a digital camera and a computer and units for converting images into pixels (10), transforming the pixels (12) into a common coordinate system and an extraction unit (14) for extracting image characteristics as well as a filter (16) for extraction of characteristics for image recognition.