Research On The Fast Specific Objects Recognition Methods In The Moving Conditions

Nowadays,object recognition technology is the hotspot and difficulty in the field of computer vision.Recognizing the objects in real time has been played an important role in the intelligent video surveillance and robot navigation.The recognition of the specific object in the mobile condition in real time is the topics of this thesis.Firstly,for the high false alarm rate and low speed of the algorithm in the past,a complete specific object recognition solution based on parallel cascade is provided.To solving the problem that the precision is low when the objects are similar,an image matching algorithm with high precision based on SVD is provided. In addition,a shot cut detection algorithm for the video clip is provided to solve the preservation problem of the surveillance video.With some experiments,we prove that the accuracy and feasibility of the method which this thesis provided.In the different conditions,the speed and accuracy of the object recognition are improved.