基于角点傅里叶描述子及空中飞机识别

Corner Based Descriptors and Recognition of Objects in the Sky

本文提出了角点傅里叶描述子,并将它作为二维目标的特征和神经网络的输入,用于构成一基于BP神经网络的飞机目标识别系统,该识别系统对旋转、位移、尺度变换具有不变性,计算机模拟结果表明该识别器对空中飞机的识别具有很高的识别率和联想能力。

We suggest a the corner based Fourier Descriptors of the object, which are taken as the features of 2-D objects and form the input of BP neural network. A multilayered perceptron architecture is used for the recognition of airplane. The scheme is invariant to rotation、translation and scale changes to object. The simulation results show that the pesented scheme gives very high recoginition accuracy and good associability.