论文部分内容阅读
目的遥感图像飞机目标的检测与识别是近年来国内外研究的热点之一。传统的飞机目标识别算法一般是先通过目标分割,然后提取不变特征进行训练来完成目标的识别。在干扰较少的情况下,传统算法的识别效果较好。但遥感图像存在着大量的干扰因素,如光照变化、复杂背景及噪声等,因此传统算法识别精度较低,耗时量较大。为快速、准确识别遥感图像中飞机目标,提出一种基于显著图和全局特征、局部特征结合的飞机目标识别算法。方法首先使用改进的Itti显著算法提取遥感图像中的显著目标;接着使用基于区域增长和线标记算法寻找连通区域来确定候选目标的数量和位置;然后提取MSA(multi-scale autoconvolution)、Pseudo-Zernike矩和HarrisLaplace特征描述子,并使用标准差和均值的比值来评估特征的稳定性,再把提取的特征结合成特征向量;最后应用支持向量机的方法完成对候选目标的识别。结果实验结果表明,本文算法检测率和识别率分别为97.2%和94.9%,均高于现有算法,并且耗时少,虚警率低(为0.03),对噪声干扰、背景影响以及光照变化和仿射变化均具有良好的鲁棒性。结论本文算法使用了图像的3种特征信息,包括MSA、Pseudo-Zernike矩和Harris-Laplace特征描述子,有效克服单一特征的缺点,提高了遥感图像飞机目标的识别率和抗干扰能力。
Purpose Remote sensing image detection and recognition of aircraft targets in recent years one of the hot spots at home and abroad. The traditional aircraft target recognition algorithm is generally through the target segmentation, and then extract the invariant features of the training to complete the target recognition. In the case of less interference, the traditional algorithm has better recognition effect. However, there are a lot of interference factors in remote sensing images, such as light changes, complex background and noise, so the traditional algorithm has lower recognition accuracy and time-consuming. In order to quickly and accurately identify aircraft targets in remote sensing images, an aircraft target recognition algorithm based on saliency map, global features and local features was proposed. Firstly, we use the improved Itti saliency algorithm to extract salient targets from remote sensing images. Then, we use the algorithm of region growth and line marking to find the connected regions to determine the number and location of candidate targets. Then, we extract multi-scale autoconvolution (MSA), Pseudo-Zernike Moments and HarrisLaplace feature descriptors. The standard deviation and the mean ratio are used to evaluate the stability of the feature. Then, the extracted features are combined into eigenvectors. Finally, the support vector machine (SVM) is used to identify the candidate objects. Results The experimental results show that the detection rate and recognition rate of the proposed algorithm are 97.2% and 94.9%, respectively, which are higher than those of the existing algorithms. The results show that the proposed algorithm has less detection time and low false alarm rate (0.03), noise interference, background effect and illumination change And affine changes have good robustness. Conclusion The proposed algorithm uses three kinds of feature information of the image, including MSA, Pseudo-Zernike moments and Harris-Laplace feature descriptors, which effectively overcome the shortcomings of single features and improve the recognition rate and anti-jamming ability of the remote sensing image of aircraft targets.