舰船监控图像拼接与识别研究

Research on mosaic and recognition of ship surveillance images

针对当前舰船监控图像拼接与识别存在的弊端,如拼接错误率高、识别正确率低等,为了提高舰船监控图像拼接与识别效果,设计了一种神经网络的舰船监控图像拼接与识别方法。首先提取舰船监控图像拼接的特征,并根据拼接关键点方向直方图建立舰船监控图像拼接模型,然后引入神经网络构建舰船监控图像识别的分类器,最后进行了舰船监控图像拼接和分类仿真模拟测试实验。相对于其它舰船监控图像拼接方法,本文方法的舰船监控图像拼接正确率得到了提升,同时本文方法的舰船监控图像识别正确率超过了90%,使得舰船监控图像的误识率大幅度减少,具有一定的实际应用价值。

In order to improve the effect of image mosaic and recognition of ship monitoring,a method of image mosaic and recognition based on neural network is designed to overcome the disadvantages of current image mosaic and recognition of ship monitoring,such as high mosaic error rate and low recognition accuracy.Firstly,the features of ship surveillance image mosaic are extracted,and the model of ship surveillance image mosaic is established according to the histogram of key points.Then,the classifier of ship surveillance image recognition is constructed by introducing neural network.Finally,the experiment of ship surveillance image mosaic and classification simulation is carried out.Compared with other methods of image mosaic for ship surveillance,the correct rate of image mosaic for ship surveillance in this method has been improved.At the same time,the correct rate of recognition for ship surveillance image in this method has exceeded 90%,which greatly reduces the false rate of ship surveillance image and has certain practical application value.