Driver posture detection method based on video and skin color area distance

Abstract

The invention discloses a driver posture detection method based on a video and a skin color area distance. The method comprises the following steps: extracting skin color areas of sampling images in aplurality of sample videos, calculating mass center coordinates of the skin color areas, converting the mass center coordinates into feature distances to represent feature values of each image, and fusing the feature values of a plurality of images corresponding to a section of video into one feature value by adopting a clustering algorithm; constructing a BP neural network, inputting the fused characteristic values and the corresponding driving posture categories as training samples into the BP neural network, and performing training to obtain a driver posture detection model; during detection, a to-be-detected video during driving of a driver is collected, a feature value of the to-be-detected video is calculated according to the method in the above steps, a calculation result serves asinput of a driver posture detection model, and the calculation result is output as the driving posture category of the to-be-detected video. The method can effectively improve the detection rate of the postures of the driver, achieves the recognition and classification of the driving behaviors of the driver, and finally achieves the real-time early warning of the operation driving process.

Description

technical field [0001] The invention belongs to the field of traffic safety, in particular to a driver posture detection method based on video and skin color area distance. Background technique [0002] The World Health Organization's "Global Status Report on Road Safety 2015" pointed out that road traffic accidents are the main factor of global population death. About 3,500 people in the world die from road traffic collisions every day. Serious road traffic accidents not only endanger the national economy, but also Bringing a heavy burden to families, improving traffic safety has become one of the top priorities in the current work of all countries. [0003] The occurrence of traffic accidents is the result of multiple factors including human-vehicle-road-environment, and researchers generally believe that more than 80% of accidents are caused by wrong driving behaviors of drivers. Most of them drive and use mobile phones. These behaviors will seriously affect the driver'...

AI Technical Summary

Problems solved by technology

[0004] As people's requirements for traffic safety are getting higher and higher, the current driver posture detection method limited to the head is difficult to meet people's requirements for safety; while the intrusive detection method has obvious limitations and is difficult to popularize; at the same time, based on the head and There are few research results in the left and right hand areas, which urgently need to be further studied and further optimized

In addition, the currently applied skin color detection method extracts the pixels of a single full image as features. The feature data is single and the dimension is too large. During the detection process, there will be problems that cannot be detected due to instantaneous area overlap, partial occlusion, etc., thus affecting the detection. Accuracy

Images