A loop detection method based on word bag model

Abstract

The invention relates to a loop detection method based on a word bag model. The lexical bag model adopts binary visual features ORB with rotation invariance and scale perception. This visual feature has the same performance as SIFT feature and SURF feature and the same computational efficiency as the visual feature composed of FAST key points and BRIEF descriptor. It is a visual feature with low computational complexity and high feature salience. The present invention employs a bag-of-words model that relies on the visual features, so that loop detection can be efficiently performed in a scenewith plane rotation and scale scaling. At the same time, the normalization method of similarity score is improved by calculating and maintaining the mean value of a normalization factor and replacingthe normalization factor when the normalization factor is abnormal. This normalization method enables the system to effectively detect loops in the event that the main body moves too fast or too slowly and steers.

Description

technical field [0001] The present invention relates to the field of visual SLAM, and more specifically, to a loop detection method based on a bag-of-words model. Background technique [0002] Loop closure detection based on the bag-of-words model is currently the mainstream practice in visual SLAM. The bag-of-words model can convert images into numerical vectors according to the distribution of visual features extracted from images in the visual dictionary, thereby realizing fast and effective comparison between images. [0003] The performance of the bag-of-words model depends on the visual features it uses. The early adoption of SIFT features and SURF features consumes a lot of time in feature extraction and matching, which aggravates the system burden. In "D.Galvez-Lopez, and J.D.Tardos, "Real-time loop detection with bags of binary words," IEEE / RSJInternational Conference on Intelligent Robots and Systems, pp.51–58, 2011" adopted by FAST The visual features composed of...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the fact that the bag-of-words model uses visual features composed of FAST key points and BRIEF descriptors, the system cannot effectively perform loop closure detection in scenes containing plane rotation and scale scaling, and avoid errors caused by abnormal normalization factors Normalization leads to the problem that the system cannot effectively perform loopback detection when the subject moves too fast or too slow and turns. The present invention proposes a loopback detection method based on the bag of words model. The technical solution adopted in the present invention is:

Images