利用支持向量机的飞机目标检测

Plane detection based on support vector machine and information fusion

为了在有干扰弹和飞机机型不定的情况下检测出天空背景中的飞机,通过自动阈值分割和腐蚀膨胀运算找出图像中的疑似区,提取其灰度方差、不变矩特征和像素比组成特征向量,应用支持向量机作为分类器对疑似区进行判别,判断其是否为飞机,选取典型的飞机图像作为正样本,选取干扰弹图像作为负样本。用不同背景和飞机机型的图像进行试验,证明该方法能很好地提出疑似区并有效地对疑似区进行检测识别出飞机目标。

In order to detect planes without fixed appearances under sky background when interfering bombs existed, automatic thresholding method and erosion/dilation arithmetic are applied to fmd out the suspicious areas of aircraft. The gray variance, invariant moments and the rate of number of pixel are extracted to constitute the image feature vector. Then the method of Support Vector Machine （SVM） is used as a classifier to detect the suspicious areas and to judge if it is a plane or not. Some typical aircrafts are chosen as the positive samples and some interfering bombs are chosen as the negative samples. Using images of different backgrounds and different aircraft appearances as test images, it is demonstrated that this method is effective on suspicious areas search and aircraft recognition.