基于HSV彩色模型的自然场景下棉花图像分割策略研究

 棉花成熟度和空间位置的识别是采棉机器人研究的关键技术，解决此问题必须对棉花图像进行分割。选取HSV模型中与亮度无关的S通道作为棉花图像的特征，排除了图像明暗变化对分割效果的影响。文中分两种情况对图像信息未缺失的棉花提出了成熟度判别的策略：正面时利用成熟棉花棉瓣的分散性和单朵棉花面积较大等特征进行判别；侧面时通过成熟棉花棉瓣相对棉荚面积比较大的特征进行识别。根据提取分割后棉花图像的几何信息，可确定棉花的重心位置。试验结果表明：该算法能很好地将成熟棉花从背景中分离出来，并较好地保存了棉花的轮廓信息。

Research on the Segmentation Strategy of the Cotton Images on the Natural Condition Based upon the HSV Color-Space Model

For the cotton harvesting robot, the degree of maturation and recognition of the position in space are the key technologies. S channel of the HSV model is chosen as the feature of cotton images, which may get rid of the effect of light. Based on two situations, a checking strategy of maturation of cotton without information lost is introduced. If it’s full-face, the ripe cotton can be judged by the dispersion and the area of one cotton, else it can be separated by the bigger area proportion of cotton pod and cotton piece. The barycenter of cotton can be ascertained by extracting geometry feature of the segmented cotton images. The experimental results demonstrate that the cotton can be separated easily from the background by the above proposed algorithm, which can well save the figure information.