基于多智能体的茶叶图像等级鉴定技术研究

摘要

Research tea image level identify, and improve the accuracy of the appraisal. The paper put forward a multi-agent tea image identification algorithm. DuoGe extraction was used with feedback function characteristics of pixels, the fuzzy feature extraction method was used for tea feature location, with feature extraction of the results as the next agent identification basis, and a small amount of fuzzy details features was complete several rounds of iterative recognition. Experimental results show that this method can effectively identify tea image levels, and good results have been achieved.%研究关于茶叶图像的等级鉴定问题,提高鉴定的准确性.针对传统的基于茶叶图像智能等级鉴定算法依靠采集到的茶叶颜色、颗粒外形饱满度等信息,靠经验完成等级鉴定,存在人为因素.当采集的茶叶图像质量较低,图像模糊,茶叶颜色、和形态信息不能很好的提取时,造成等级鉴定不准确的问题.为了解决上述问题,提出了多智能体的茶叶图像等级鉴定算法.通过提取多个带有反馈功能的像素特征,通过模糊特征提取的方法进行茶叶特征定位,用特征提取的结果作为下一个智能体识别的依据,利用少量模糊的细节特征完成多轮选代识别.实验表明,改进方法能够对茶叶图像等级进行有效的鉴定,取得了比较好的效果.