论文部分内容阅读
提出利用线性判别函数，ｄ（ｘ）＝Ｗ′ｘ，对轰炸机和战斗机两类目标进行旋转不变识别。每一目标有７２个旋转模式，其中９个为训练样本，６３个为测试样本，在每个目标的训练样本中随机取一个相减作为线性判别函数的系数（Ｗｉｊ）的初始值，然后利用误判的训练样本优化判别函数的权值以提高函数的正确识别率。为便于光电实现，线性判别函数的权值系数只有２、１、０、－１、－２五个值。计算机模拟表明利用此判别函数进行分类的正确识别率达９５．８％。文中还提出一个实现此线性判别函数的光电混合识别系统。
In this paper, the linear discriminant function d (x) = W’x is proposed to detect the invariant rotation of bomber and fighter. Each target has 72 rotation modes, of which 9 are training samples and 63 are testing samples. Randomly subtract the initial value of the coefficient (Wij) which is a linear discriminant function in each target training sample, and then use Misjudged training samples optimize the weight of the discriminant function to improve the correct recognition rate of the function. In order to facilitate the realization of photovoltaic, the linear discriminant function weight coefficient only 2,1,0, -1, -2 five values. Computer simulation shows that using this discriminant function to classify the correct recognition rate of 95.8%. The paper also presents a photoelectric hybrid recognition system to achieve this linear discriminant function.