Indoor large-scene pedestrian fingerprint positioning method based on monocular vision

Abstract

The invention relates to an indoor large-scene pedestrian fingerprint positioning method based on monocular vision. Firstly, under the condition that the position of a monocular camera is fixed, a positioning scene is divided based on precision requirements, fingerprint information is collected, then a to-be-positioned image is collected through the camera, pixel coordinates of the upper left corner and the lower right corner of a human face are obtained through human face recognition, the area of the human face is calculated, matching is conducted, and coordinate candidate points of a person are obtained; and finally, a distance value is calculated between the left upper corner coordinate of the human face and the left upper corner coordinate in the database, wherein the coordinate corresponding to the minimum distance value is personnel position information. A fingerprint matching method is adopted for positioning, the difficulty that initial information of a positioning object needs to be known based on a photo inter-frame pose resolving method is eliminated, position information can be directly obtained, the method can be directly applied to existing hardware equipment, the utilization rate of existing resources is fully improved, and the positioning requirement of a large scene is met with lower cost.

Description

technical field [0001] The invention belongs to the technical field of indoor positioning, in particular to a monocular vision-based pedestrian fingerprint positioning method in a large indoor scene. Background technique [0002] Affected by the occlusion of the building itself and the internal structure, the satellite signal will be weak or even missing in the indoor environment, making the means based on GNSS positioning unable to be used for indoor positioning. The current indoor pedestrian positioning solutions are mainly based on Bluetooth, WIFI, vision and other technical means. [0003] In a large scene, in order to obtain high-precision positioning information, the positioning scheme based on Bluetooth and WIFI needs to deploy a large number of hardware devices to divide the scene and obtain better positioning accuracy with high-density hardware distribution. However, the positioning cost and equipment Maintenance costs also increase. The positioning scheme based o...

AI Technical Summary

Problems solved by technology

The positioning scheme based on binocular vision has a very mature application in the field of industrial inspection, but the positioning distance is relatively short, with only a decimeter-level range of action. In large scenes, it is necessary to use a high-resolution lens to identify people, and the camera’s Trigger synchronization has extremely high requirements, and it is difficult to apply in practice

The positioning scheme based on monocular vision can solve the continuous relative change information of the pedestrian's pose by taking the imaging pictures before and after the pedestrian, but it cannot obtain the absolute position coordinates, and it needs to combine the known initial state information to complete the absolute positioning. The randomness of the camera field of view appears contradictory

In summary, the current indoor positioning methods cannot complete the absolute position positioning of pedestrians in large-scale scenes at low cost.

Images