基于视觉注意机制的交通路标检测方法

提出一种基于视觉注意机制的交通路标检测方法,该方法在图像灰度变换的基础上,引入自下而上的视觉注意模型,提取图像的初级特征,构造相应的显著图,根据总显著图,检测出图像中的显著区域,定位路标在图像中的位置。对多幅实时路况图像的实验结果表明,该方法能够适应户外自然环境检测,具有较高的准确性和实时性。

Approach to waymark detection based on visual attention mechanism

This paper proposed a new approach to waymark detection based on visual attention mechanism. This approach introduced the bottom-up human visual attention model after the image grey scale transformation, extracted low-level features of the image and constructed the saliency map. According to the whole saliency map, it detected the salient regions in an image, fixed the location of waymark in an image. The experimental results on multiple real-time traffic images demonstrate this approach can adapt to the test in an outdoor natural environment and be of high accuracy and real-time.