Study On The Method Of Visual Robot Environment Cognition

At present, the mobile robot based on computer vision is playing an important role in various fields including life and labor. With the development and improvement of all kinds of sensors and sensing systems, the SLAM(Simultaneous Localization And Mapping) problem has been proposed for mobile robot in recent years. This kind of problem can be simply described as placing a robot in an unknown space environment, the sensors sense the surroundings by its own movment. The robot will reconstruct the map of the space, and judge the location itself real-time. Through the research of this kind of problem, it can make the vision robot have the same ability as person in region cognition. The vision robot can draw its own trajectory without GPS module to build the map of unknown area. Thus,solving this problem has lots of appliaction value and developing foreground in the fields including national defense and prospecting,and has become an hot issue in mobile robot research. In order to make the research of SLAM problem based on binocular vision mobile robot completely,this paper mainly from the following four aspects of research.1. Proposing a calibration method of binocular camera, gives two kinds of 3D reconstruction algorithm based on binocular image. Reconstruct a set of binocular image by the calibration binocular camera, and compare the excellent degree of each algorithm.2. Due to the images captured by binocular cameras will be stored together in a mess. The method of artificial classification can lead to a huge amount of work. A color image classification algorithm based on LDAIC(Latent Dirichlet Allocation for Image Classification)model is proposed to classify the multi-scene images.3. A moving object tracking technique based on Kalman filter algorithm is proposed.With the movement of objects, the bonocular camera collects the image sequences based on the scene.Further, a joint analysis of binocular images is presented, which is based on the binocular vision of Kalman filter object tracking algorithm.4. Based on binocular vision based Kalman filtering algorithm by the above mentioned, a moving object orientation judgment method is presented. This method can judge the direction of motion of the object. According to the judgment of this direction, a solution of SLAM problem based on binocular vision is proposed.