首先对自适应遗传算法的变异算子进行了改进,对单点变异算子与双点变异算子的结合能有效地改善局部收敛进行了验证,然后提出了一种新的用自适应遗传算法分割图象的方法,并与传统的Otsu方法、灰度差直方图法和基于熵的方法作了比较。实验表明,该文的算法可保留图象的大部分信息,对一些复杂图象的处理能得到很好的处理结果,同时本文算法在时间上还有很大的优势.

Firstly, an improvement was made on mutation of adaptive genetic algorithm (AGA). The combination of one-point mutation and two-point mutation could prevent AGA from converging to the local optimum. Secondly, a new method for image segmentation using the improved AGA was put forward. The method of Otsu, the method of histogram based on gray difference and the entropy-based methods were compared with this algorithm. The improved AGA was proved to give attention to the main information by experiments. Good results can be achieved from some images. And much less time was used by the algorithm.