一种基于ROI的红外舰船目标定位方法

An Object Localization Algorithm Based on ROI for Ship IR Image

提出了一种基于感兴趣区域ROI(Regions of Interest)的红外舰船目标定位方法,通过改进的Itti模型提取包含目标的感兴趣区域,实现目标定位。首先应用小波变换代替Itti模型的高斯滤波生成图像多尺度金字塔,并用center-surround算子提取多尺度的视觉差异,再将生成的视觉特征图进行归一化并线性组合,生成显著图,最后运用交替式有效子窗口搜索算法A-ESS(Alternating Efficient Subwindow Search)定位目标区域。实验结果表明:该方法能准确定位出目标区域。

An object localization algorithm based on ROI for ship IR image was developed. llae KOI with target was extracted by the improved Itti model. Firstly, the multi-scale pyramid was created by the wavelet transformation. After center-surround differences between different scales of the pyramids being calculated, some feature maps were obtained. Then these feature maps were combined into a saliency map with normalization and linear-fusion. Finally, object location was obtained by A-ESS in saliency map. Extensive experiments validate the good performance of the approach.