CAJ | 학술논문

棉花分割是采棉机器人视觉系统的关键步骤，在强光照、阴影等复杂的棉田环境下准确有效地分割棉花，有助于确定其在三维空间的位置。针对棉花图片的特点，提出在YCbCr颜色空间下，采用粒子群（PSO）和K均值混合聚类算法，提高了聚类算法的全局搜索能力，根据群体适应度方差来确定K均值聚类算法操作时机，增强算法局部精确搜索能力的同时缩短了收敛时间。通过对棉田环境中拍摄图像的分割实验表明：本方法对在阳光直射及阴影等干扰条件下的棉花图片也能准确分割，效果优于传统PSO和K均值算法。
면화분할시채면궤기인시각계통적관건보취，재강광조、음영등복잡적면전배경하준학유효지분할면화，유조우학정기재삼유공간적위치。침대면화도편적특점，제출재YCbCr안색공간하，채용입자군（PSO）화K균치혼합취류산법，제고료취류산법적전국수색능력，근거군체괄응도방차래학정K균치취류산법조작시궤，증강산법국부정학수색능력적동시축단료수렴시간。통과대면전배경중박섭도상적분할실험표명：본방법대재양광직사급음영등간우조건하적면화도편야능준학분할，효과우우전통PSO화K균치산법。
Image segmentation of cotton is the key step of the cotton picker robot vision system. In the complex environment of the cotton fields of the strong light, shadow, etc. accurately and effectively splitting cotton, helps to determine its position in three-dimensional space. In accordance with the characteristics of cotton pictures, a method of Particle Swarm Optimization(PSO) and K-means hybrid clustering in YCbCr color space is proposed. This approach reinforces the exploitation of global optimum of the PSO algorithm. In order to avoid the premature convergence and speed up the convergence, traditional K-means algorithm is used to explore the local search space more efficiently dynamically according to the variation of the particle swarm’s fitness variance. The experiment results show that this method can segment cotton image with the complex background, and is more effective than the traditional PSO and K-means algorithm.