四元数聚类算法在苹果图像识别中的应用

Application of Quaternions Clustering Algorithm in Apple Image Identification

针对自然环境下苹果图像识别问题,提出在RGB颜色空间中,利用彩色图像像素点四元数表示和四元数旋转特性,改进K-means聚类算法,完成苹果目标的分割识别。该算法能有效区分苹果目标和背景,识别精度高,不但可以应用到成熟红苹果分割识别,还能应用于青苹果分割识别。实验表明,该四元数聚类算法可以取得较好的苹果识别效果。

Aiming at the apple image in the natural environment,an improved K-means clustering algorithm based on quaternion representation and quaternion rotation is proposed to identify apple in RGB color space.The algorithm can effectively distinguish the apple target from image background with high recognition accuracy.It can not only be applied to identification of mature red apple,but also to identification of green apple.Experimental results show that the quaternion clustering algorithm can effectively identify apples in color image.