Seat belt not-wearing detection method based on mixed multi-scale deformable component model

Abstract

The invention discloses a seat belt not-wearing detection method based on a mixed multi-scale deformable component model, which comprises a deformable component model training process and a process for seat belt detection by using the deformable component model. The process for seat belt detection by using the deformable component model comprises steps: a picture is acquired through a picture acquisition device; the deformable component model is used for vehicle window detection; human upper body detection is carried out in the detected vehicle window area; and seat belt detection is carried out on the detected human upper body area.

Description

technical field [0001] The invention relates to the technical field of object detection, in particular to a detection method for not wearing a safety belt based on a hybrid multi-scale deformable part model. Background technique [0002] Computer vision technology is a technology that studies how to use computers and related equipment to simulate biological vision. The image or video is collected and processed by imaging equipment such as a camera to obtain the three-dimensional information of the corresponding scene, and then the computer replaces the brain to complete the processing and understanding. The technology involves multiple disciplines, including image processing, pattern recognition, image analysis, and image understanding. At present, computer vision technology has been widely used in various fields, such as medical image processing, video surveillance, electronic checkpoint, virtual reality, intelligent transportation, etc. [0003] As the most effective pro...

AI Technical Summary

Problems solved by technology

[0015] The above scheme has the following disadvantages: when positioning the vehicle window area on the basis of the license plate area positioning, the correct detection rate of the vehicle window largely depends on the accurate positioning of the license plate, but the license plate positioning effect is less affected by the license plate inclination angle and the size of the rectangle. Large, and the area of ​​the license plate area is relatively small compared to the whole image, which consumes a lot of detection time

Although some researchers use the method of detecting the left and right corners on the top of the car window to locate the seat belt area, due to the influence of light, the left and right corners on the top of the car window will be blurred, and it is impossible to determine whether there are occupants in the window. Detecting seat belts, there will be many false detections

In addition, due to the influence of the upper edge area of ​​the car window, the windshield of the car window, and lighting, sometimes the face image is often partially or completely blocked, resulting in incomplete, side-faced, blurred and other low-quality situations in the face image. Moreover, The maturity of the face detection algorithm is not enough, and the detection of seat belts based on the method of face detection will lead to missed detection

Use the geometric features of the straight lines on both sides of the seat belt to detect the seat belt. Since the angles between the two lines and the distances between them are different, the length of the lines cannot be unified, and there are many cases where the straight lines in other areas are misdetected as seat belts.

At the same time, the straight line detection algorithm is greatly affected by illumination and geometric deformation, and there are too many parameters in the algorithm process. Different samples need different parameters to be detected, and the robustness is not strong.

Images