改进的Otsu理论在图像阈值选取中的应用

提出了一种基于GA优化的Otsu理论进行图像阈值选取的新方法.利用传统的Otsu理论进行图像阈值选取,计算量较大,准则函数不一定单峰,不适于最佳阈值的求取.遗传算法理论为一种全局搜索方法,它自适应地控制搜索过程以求得最优解,从而可克服Otsu方法的不足,有利于计算机视觉的后续处理.文中将遗传算法和Otsu理论进行了有机结合,实现了图像阈值自动选取,且大大降低了计算量.实验结果表明该算法不仅提高了分割质量,而且缩短了寻优时间,从而说明了该算法的有效性、正确性.基于改进Otsu优化的模糊算子理论的提出,解决了模糊算子中关键参数确定困难的问题,实现了其参数的自适应获取,并将该理论应用于图像增强中,从而有效地消除了图像的模糊和噪声干扰.

Automatic Selection of Image Threshold Based on Improved Otsu

A theory of Otsu optimized by GA is improved. The traditional Otsu is inefficient and the peak value of the rule function may not be exclusive. The GA , which confirms the key operators adaptively, conquers the shortage of Otsu theory. The method based on genetic algorithms and Otsu theory realizes automatic selection of image threshold, and reduces the operation. The result of experiments show that the method is not only of higher segmentation quality but also of higher computational speed. So, it proved that the algorithm is right and efficient. The theory of fuzzy operator optimized by Otsu, which confirms the key operators adaptively, is applied to image enhancement. The fuzziness and the noises are eliminated efficiently.