基于改进ITTI模型的SAR图像目标检测

Target Detection of SAR Images Based on Improved ITTI Model

合成孔径雷达作为主动式雷达,具有全天时、全天候、高分辨、可侧视成像等优点,在军事和民用领域得到了广泛的应用。本文将视觉注意机制引入SAR图像目标检测中,针对SAR图像特征,改进传统视觉注意计算模型,并研究了基于视觉显著性的SAR图像目标检测方法。归纳总结了现有的视觉注意计算模型,深入研究了传统ITTI模型中基于这种自底向上的视觉注意模型对不同的SAR图像进行目标检测实验;针对SAR图像具有丰富纹理信息的特点,将纹理特征引入视觉注意模型的初级视觉特征提取阶段,提出了一种合理分配纹理特征与强度特征权重的改进方法,并采用该方法对SAR图像进行了目标检测。实验表明,本文的方法能避免ITTI模型存在的漏检现象,并一定程度上提高了检测区域的完整性。

Synthetic Aperture Radar （SAR） is the active radar. As SAR can work in nearly all weather conditions, it has been widely used in many fields. This thesis studies the visual attention model and then introduced the visual attention model to the target detection of SAR images. In view of the SAR image characteristics, we improved the traditional visual attention model, and studied SAR image target detection method based on visual significance. The main contributions of this thesis are follows. Mechanisms of human visual attention were analyzed and we introduced the mechanisms in both areas of neurobiology and psychology. We summarized the existing calculation model of visual attention. We further study of the ITTI model, and we use the bottom-up model to detect the different targets in different areas in the SAR images. According to the characteristics of SAR images have rich texture information, we proposed a improved visual attention model, the model through the extraction of image texture features, and we use the improved model to detect the ROI information in SAR images. The experimental results show that the model generate better saliency map, has fast calculation speed, accurate positioning, the outline is clear.