Image deblurring method based on camera positioning

Abstract

The invention discloses an image deblurring method based on camera positioning. The method comprises two stages: a blurred image deblurring stage and a deblurred image feature extraction stage. In theblurred image deblurring stage, the depth information of the scene is obtained by using the depth image of the blurred image of the current frame of the camera, and then the three-dimensional coordinates of the scene point are obtained. And camera motion information including translation and rotation is obtained by using an inertial measurement unit (IMU). And a blurred kernel of the selected block is calculated by using the data, and deconvolution operation is performed by using the blurred kernel to obtain a deblurred image. A deblurred image feature extraction stage: ORB feature extraction is performed on the deblurred image, and a subsequent SLAM process is performed by using the extracted features. According to the method, only the selected block is deblurred, so that the calculation amount is reduced to a certain extent, and the operation speed is increased.

Description

technical field [0001] The invention belongs to the field of computer vision, especially for motion blur caused by camera movement within exposure time, and in particular relates to an image deblurring method based on camera positioning. Background technique [0002] In visual SLAM, blurring occurs in captured images due to camera shake during the exposure time. The presence of motion blur makes it difficult to perform data association on reconstructed landmarks and reconstruct new features, leading to failure of SLAM system localization or reconstruction. [0003] As an important direction in the field of image restoration, image deblurring has a wide range of applications in real life. In SLAM, the image can provide more features after deblurring, so that the system can better perform data association and reconstruct new features, and can successfully continue the SLAM process in subsequent frames, improving the robustness of visual SLAM to motion blur . [0004] High-q...

AI Technical Summary

Problems solved by technology

[0004] High-quality methods for motion blur removal have been developed in recent decades, but most are computationally intensive, making it difficult to use these methods to restore images for visual SLAM

Based on the statistical information of natural image gradients and the variational Bayesian method, Fergus et al. have achieved good results in the restoration of a single image, but the parameter estimation results in a very slow speed, and the image restoration results of some large blur kernels not stable enough

On this basis, QiShan et al. analyzed the generation of ringing effects, and proposed global priors and local priors, which achieved better results in the accuracy of blur kernel estimation and the suppression of ringing effects. effect, but the convergence speed of the algorithm is not fast enough, and the setting of parameters has a great influence on the result, and the time complexity is large

Images