TECHNIQUE FOR OBJECT ORIENTATION DETECTION USING A FEED-FORWARD NEURAL NETWORK

摘要

A TECHNIQUE FOR OBJECT ORIENTATION DETECTIONUSING A FEED-FORWARD NEURAL NETWORKAbstractThe present invention relates to a technique in the form of an exemplarycomputer vision system for detecting the orientation of text or features onan object of manufacture. In the present system, an image of the featuresor text is used to extract lines using horizontal bitmap sums, and thenindividual symbols using vertical bitmap sums, using thresholds with eachof the sums. The separated symbols are then appropriately trimmed andscaled to provide individual normalized symbols. A Decision Modulecomprising a Feed-Forward Neural network and a sequential decisionarrangement determines the "up", "down" or "indeterminate" orientation ofthe text after a variable number of symbols have been processed. Thesystem can then compare the determined orientation with a database tofurther determine if the object is in the "right-side up" "upside-down" or"indeterminate" orientation.