改进二维OTSU和自适应遗传算法的红外图像分割

Infrared Image Segmentation Based on Improved Dimensional OTSU and Genetic Algorithm

为提高电路板热红外图像分割效果,解决电路板故障诊断的问题,提出了一种将自适应遗传算法和改进二维OTSU算法相结合的新算法。自适应遗传算法采取快速排序法,利用遗传代数的变化改进交叉、变异概率;根据邻域像素点与中心像素点之间的距离对二维OTSU邻域均值算法的比例系数进行加权,判断噪声点并对热红外图像进行降噪处理;引入类内方差法,对阈值进行优化。利用了遗传算法的并行性和强大的空间搜索能力,提高了二维OTSU的阈值查找速度,提高了热红外图像的分割效率。实验结果表明,该算法提高了热红外图像的分割准确度,有一定的应用前景。

In order to improve the image segmentation of circuit board, and solve the problem of circuit board fault diagnosis, a new algorithm combines adaptive genetic algorithm and improved two-dimensional OTSU algorithm was proposed, a fast sorting method, in accordance with changes in the genetic algebraic, crossover and mutation probability was improved,＂ according to the distance between the center and neighboring pixels pixel for the neighborhood average algorithm to weight the scale factor, on the basis of judgment on noise point of thermal infrared image to reduce noise; and introduced into the class variance method to solve the equation for the threshold to optimize. The new algorithm uses a genetic algorithm parallelism and powerful spatial search capability, improving dimensional OTSU threshold search speed and improving the efficiency of thermal infrared image segmentation. Experimental results show that the algorithm improves the accuracy of thermal infrared image segmentation, there are certain applications.