Rainy day traffic video saliency detection method based on double-channel visual mechanism
The invention discloses a rainy day traffic video saliency detection method based on a double-channel visual information processing mechanism, which is applied to the technical field of computer vision and aims at solving the problem that a rainy driving task is more difficult due to many factors su...
Saved in:
Main Authors | , , , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
18.05.2021
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses a rainy day traffic video saliency detection method based on a double-channel visual information processing mechanism, which is applied to the technical field of computer vision and aims at solving the problem that a rainy driving task is more difficult due to many factors such as low visibility, rainwater, pedestrians and umbrellas in cloudy and rainy weather. According to the method, firstly, an eye movement behavioral experiment is designed, and a rainy day traffic video data set containing eye movement fixation points of 30 drivers is collected under driving of a driving task; based on the inspiration of a visual information processing mechanism, a dual-channel video saliency detection model is proposed, space and time feature information is extracted from static and dynamic channels in the model respectively, and then features are fused, so that a traffic video saliency map in rainy days is effectively predicted, and the method has important significance for the development of an |
---|---|
Bibliography: | Application Number: CN202110140442 |