基于道路场景的天气识别与目标提取方法研究

Road Scene Research Based on Weather Detection and Target Extraction

针对晴、阴、雾天环境下道路场景中运动目标提取问题,探讨基于特征量融合的天气类型判别算法,借助不同天气下的图像特征信息研究运动目标提取算法。基于不同天气下多种场景的视频图像序列,实验结果表明,获得的天气类型判别算法能较好地依据图像的历史信息判别天气的类别;目标提取结果表明,获得的目标提取算法能有效消除噪声,并且提取运动目标的效果较好。

To solve the problem of extracting moving object in road scene in sunny/cloudy/foggy weather,a weather type discrimination algorithm based on feature fusion was discussed. Under discussion,a moving object extraction algorithm was proposed by studying the multi-feature image information under different kinds of weather. The experimental results show that based on video-frequency image sequence with multiple scenes under different weather types,the weather-type discrimination algorithm can determine the weather types according to the historical information of an image on one hand. On the other hand,the moving object extraction algorithm can eliminate effectively the background noises of images and extract the moving target included in such images.