一种混合遗传算法在彩色图像分割中的应用

磨粒图像分割是磨粒识别的基础,其分割精度直接影响到磨粒自动识别的精度。在磨粒图像RGB彩色空间,选择其3个正交彩色特征,应用遗传算法和类间最大方差法对其分割,依据F isher判别函数选择其最佳特征与最佳阈值,实现了磨粒彩色图像的分割。大量磨粒图像分割实践证明,该分割方法具有分割精度较高、计算速度快、自适应性强等特点,是铁谱磨粒彩色图像自动分割较理想方法之一。

Research on Auto-segmentation of Ferrograph Wear Particle Color Image

The wear particle image segmentation is the base of wear particle recognition,and its segmentation precision affects on wear particle recognition precision directly.In the RGB color space of wear particle color image,three vertical-intersectant character were chosen,its images were segmented using genetic algorithm and maximum-variance algorithm in all classes,and its best character and best threshold were chosen by Fisher differentiating function.A number of practices prove that the approach has high segmentation precision,fast computation velocity and self-adaptive ability,and it is a ideal method of wear particle color image auto-segmentation.