A 3D pose estimation method of image vehicle based on fine CAD model

Abstract

The invention discloses an image vehicle attitude estimation method based on a fine three-dimensional model. The method comprises steps: first, the position of the vehicle being detected from the image, and initializing the vehicle three-dimensional posture parameters, rendering the vehicle three-dimensional model to the image plane using the current posture parameters, extracting the contours ofthe image vehicle and the model vehicle respectively, using the matching error between the contours to construct the energy function, and adopting the Gauss-Newton algorithm to optimize the posture parameters, solving the energy function minimization problem, and obtaining the final result. In contour matching, the improved random sampling consistency (RANSAC) algorithm is used to fit the contourof the curve model with piecewise straight lines, and then the contour is matched with the real vehicle contour. The invention can accurately and robustly recover the three-dimensional posture parameters of the vehicle relative to the camera in the surveillance video in the complex surveillance scene, which is of great significance to the understanding and automatic processing of the surveillancevideo.

Description

technical field [0001] The invention belongs to the technical field of computer vision, in particular to a method for estimating a three-dimensional attitude of an image vehicle based on a fine CAD model. Background technique [0002] With the continuous advancement of the construction of a safe city, more than 30 million surveillance cameras have been deployed in my country. These cameras generate a large amount of surveillance video data every day. Therefore, the requirements for automatic processing of these surveillance videos are also getting higher and higher. In the surveillance video, driving vehicles are one of the key objects of concern. Accurately estimating the three-dimensional attitude parameters of vehicles is the basis of automatic processing of surveillance video, which can be widely used in traffic monitoring, road law enforcement, security assurance, and surveillance video compression processing. [0003] The difficulty of vehicle attitude estimation in s...

AI Technical Summary

Problems solved by technology

This method is one of the methods with the best comprehensive effect in the field of pose estimation at present, but because only the outer contour of the target is used, the internal contour information of the target is not fully utilized, which makes it not robust enough in the case of the target being occluded and complex lighting

[0005] Existing methods have problems such as low accuracy of attitude parameter estimation and poor robustness in complex environments.

Images