摘要
为实现海事动态视频监测中海面目标的自动检测,提出基于小波域视觉注意选择机制的海面目标快速检测算法。根据人类视觉观测特点,首先利用提升小波变换在小波域建立了双尺度视觉选择注意模型,然后在粗分辨率低频子带上分别利用相位谱法和梯度法建立视觉显著图,并对两者进行有效融合形成综合视觉显著图,最后通过小波反变换得到原始高分辨率图像的视觉显著图,并由此实现海面目标区域的提取。实验结果表明:该算法能够快速、准确地实现海面目标的自动检测,可用于基于海洋浮标的海事智能监测。
In order to achieve the automatic detection of sea surface objects for maritime monitoring from dynamic video,this paper presented a fast algorithm to detect sea surface targets based on the visual selective attention mechanism in wavelet domain.According to the characteristics of human visual observation,a two-scale visual attention model was first established in the wavelet domain by using lifting wavelet transform.Then two visual salience maps were generated by employing the phase spectrum method and gradient based method on the low-pass subband of wavelet domain with coarse resolution respectively,and a synthetic visual salience map was obtained by effective combination of the both obtained saliency maps.Finally,the high resolution visual saliency map of original image was generated by inverse wavelet transform,and the sea surface object regions were extracted from the final saliency map.The experimental results show that the proposed algorithm can detect the maritime targets quickly and accurately,so it can be used for maritime buoy-based intelligent monitoring.
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2015年第4期110-115,共6页
Periodical of Ocean University of China
基金
国家高技术研究与发展计划项目(2006AA09Z237)资助
关键词
视觉注意机制
视觉显著性
海面目标检测
小波变换
visual attention mechanism
visual salience
sea surface object detection
wavelet transform
