IC芯片视觉检测中快速图像匹配定位

Fast Image Matching Location in IC Vision Measurement

利用IC芯片定位图像特征,提出一种用其投影特征的基于优势遗传自适应遗传算法(AGA)的快速匹配算法(P-AGA)。设计基于图像灰度投影特征的匹配规则,使算法适应于一定的噪声、缩放、旋转和变形图形,而且运算量小。对图像采用2×2重采样,使运算量降低3/4。匹配过程是在二维图像空间的随机寻优,提出优势遗传的AGA,用匹配距离作为适应度函数,并自适应改变交叉变异概率进行匹配寻优,快速准确求出全局最优匹配点。实验表明,该算法比标准遗传算法(SGA)快约1.9倍,准确度提高约22.5%,有效提高匹配速度和准确度,有很强的抗噪声能力,鲁棒性好。

To locate image＇s feature by using IC,a fast image matching algorithm using image projection feature based on superiority inheritance adaptive genetic algorithm（P-AGA） is presented design of the matching rulesis Design matching rule based on image gray projection feature, so the algorithm suits some noise, scaling, revolution and deformed images, and its computation capacity is low. Images are resampled by 2×2 by using the algorithm, and the computation capacity can be reduced to 1/4. Image matching is a random search problem in 2-D image space. Adaptive genetic algorithm based on superiority inheritance is presented, which regards match-distance as fitness function,and varies the probabilities of crossover self-adaptively. The experiments demonstrate that P-AGA is faster about 1. 9 times than standard GA（SGA）,and its accuracy is increased about 22.5%. So P-AGA improves matching speed and accuracy efficiently, and it has high anti-noise ability and robust.