Binocular visual sense mileage calculating method based on image gradient combined optimization

Abstract

The invention discloses a binocular visual sense mileage calculating method based on image gradient combined optimization; the method comprises the following steps: preprocessing an inputted binocularvideo, and calculating a binocular video frame disparity map so as to obtain a tolerance distance; zooming to form multi-scale images and forming a pyramid model, and calculating features to obtain aseries of feature images; using a Kalman filter to process the image according to the tolerance distance, and predicting and estimating a camera attitude motion process, wherein a camera motion modelis built in the Kalman filter; using a binocular visual sense navigation algorithm based on gradient to calculate the current frame accurate camera attitude; using the current frame camera attitude to update the camera motion model in the Kalman filter. The invention proposes the optimization algorithm combining two gradient expansions, and creatively uses the image gradient as features, thus effectively preventing outdoor illumination change influences; the method uses a plurality of key frames so as to optimize the camera attitudes, is good in performance, and can simultaneously carry out dense three dimensional reconstructions.

Description

technical field [0001] The invention relates to the field of computer vision and related technologies of image processing, in particular to a binocular vision mileage calculation method based on joint optimization of image gradients. Background technique [0002] After years of exploration and development, computer vision has been applied in many fields such as robotics, autonomous driving, digital entertainment, medical health, security monitoring, and military technology. Visual odometry is not only an important part of the visual navigation system for robots and unmanned vehicles, but also an important research branch in the fields of computer vision and robotics. This invention is dedicated to proposing a fast and effective binocular visual odometry method. Visual odometry is the research foundation and premise of target positioning, attitude estimation, map drawing, scene understanding and other topics. An excellent visual odometry calculation method can provide strong ...

AI Technical Summary

Problems solved by technology

However, the main defect of the direct method is the small motion assumption of the camera and the tendency to converge to a local optimum, which makes the performance of this type of method weaker than the strategy based on feature point matching in the KITTI visual odometry test.

[0005] In the current visual navigation KITTI evaluation platform, the direct visual navigation method based on dense observations fails to achieve ideal results. The main reasons can be attributed to the following points: 1) Image alignment based on the Lucas-Kanade framework [3] The theoretical convergence area is limited, and it cannot cope with the large-displacement scene sampled at 10Hz in the data set, which requires a very good motion estimation method to compensate; 2) The KITTI data set is derived from actual driving scenes, covering different driving environments including cities, villages, and highways , directly using the pixel gray value and the affine lighting model cannot cope with the impact of complex environmental lighting conditions such as road reflection, tree shade, and building shadows; 3) imperfect camera calibration and the accuracy of stereo vision matching algorithms directly affect the accuracy of visual navigation results , the current stereo vision matching evaluates the accuracy of disparity estimation within the coverage, while visual navigation requires accurate point-to-point disparity estimation, and the algorithm that ranks in the forefront of stereo vision evaluation fails to effectively improve navigation accuracy

Images