Abstract: In order to separate out the petals from flower images, a double threshold color image segmentation method is proposed in this paper based on clonal selection algorithm and maximum classes square. The color of images is quantified in the HSV color space and the color of flowers is calculated automatically. The component operators make a pretreatment to images according as color category. Clonal selection algorithm analyses matrix got from the last step and counts out the global optimal threshold. Experimental results show that, the proposed method realizes automatic segmentation. This algorithm is effective, shortens computing time and is feasible.

Key words: flowers, image segmentation, clonal selection, color quantization, OTSU, double threshold

摘要： 为了准确分割出花卉图像中的花朵，针对这类图像的特点，提出了一种基于克隆选择算法和OTSU（最大类间方差）法的双阈值彩色图像分割法。在HSV颜色空间中对图像进行颜色量化，计算出花卉的大致颜色，依据颜色种类对图像进行分量算子预处理，运用克隆选择算法对上一步中得到的结果矩阵自学习，得到全局最优的阈值，从而实现图像分割。实验结果表明，该方法无需人工干预，分割效果良好，且大大缩短了计算时间，具有一定的实用性。

关键词: 花卉, 图像分割, 克隆选择, 颜色量化, 最大类间方差, 双阈值