基于色彩空间的航拍油菜花图像分割

Color space-based aerial rapeseed flower image segmentation

针对花期识别过程中油菜花航拍图像分割效果不理想的问题提出一种基于色彩空间分割油菜花图像的方法。首先在Lab空间利用K-means算法对油菜图像进行初次处理,然后在HSV空间使用颜色分割算法对聚类结果进行二次处理,最后把分割出的油菜花映射到RGB空间,以期完成油菜与油菜花的有效分割。选取了60幅不同的航拍油菜图像进行分割实验。结果表明采用此方法不仅能够较好地分割出油菜花,而且还能有效地避免光照的影响。此实验结果能为后期识别油菜花花期生长规律的研究提供参考。

Aiming at the problem of unsatisfactory segmentation effect of aerial image of rapeseed flowers in the process of florescence recognition.This paper proposes a method combining K-means algorithm and color segmentation algorithm to segment rapeseed image.Firstly,the K-means algorithm was used to first process the rapeseed image in Lab space.Then,the clustering results were processed once again in HSV space using color segmentation algorithm.Finally,the segmented rapeseed was subjected to morphological treatment to complete the effective segmentation of rapeseed and rapeseed flowers.Sixty different aerial rapeseed images were selected for segmentation experiments.The results show that this method can not only segment rapeseed well,but also effectively avoid the influence of illumination.The results of this experiment can provide reference for the later study of the flowering period of rapeseed.