遗传算法在小目标图像分割中的应用

将遗传算法用于小目标图像分割 ,提出了目标在图像中所占比例的可能范围参数 ,结合P tile法和直方图熵法 ,得到了一种新的自适应目标分割方法。该方法克服了传统P tile法要求已知目标所占确切比例的缺陷 ,并利用了遗传算法能自动在搜索空间内快速寻优的特点 ,而且可以推广到任意大小目标的图像分割问题上。试验结果表明 ,该方法具有良好的分割质量 ,运算速度提高了 2 1 5 %。

Application of the Genetic Algorithm in Small Object Segmentation

An adaptive approach to small object segmentation based on genetic algorithms is proposed. A new parameter "scale of the object area's percentage" is introduced in this method. The method overcomes the P tile method's defect of requiring the exact percentage of an object area, and meanwhile makes effective use of the small object's characteristics. Based on the genetic algorithm, the new approach can automatically locate the optimal threshold in the search space, and is possible to be extended to any objects. Experimental results indicate that the proposed algorithm achieves a good segmentation quality and improves the computational efficiency by 21 5%.;