An image gradient combined optimization-based binocular visual sense mileage calculating method, comprising: preprocessing an input binocular video, and calculating a disparity map of binocular video frames so as to obtain a measure distance; zooming to form multi-scale images and forming a pyramid model, and calculating features to obtain a series of feature images; using a Kalman filter to process the images according to the measure distance, and predicting and estimating a camera attitude motion course, a camera motion model being built into the Kalman filter; calculating the accurate camera attitude of a current frame by using a gradient-based binocular visual sense navigation algorithm; and using the camera attitude of the current frame to update the camera motion model in the Kalman filter. The described method provides an optimization algorithm combining two gradients, and creatively uses an image gradient as a feature to effectively prevent the influence of a change in outdoor brightness. Camera attitudes are optimized by referring to a plurality of key frames, thereby obtaining a real-time binocular visual sense mileage calculating method that has good performance and is capable of carrying out dense three-dimensional reconstruction.
展开▼
