应用叶片图像分割与特征融合的复杂背景植物识别方法

针对复杂背景的存在性,通过图像分割处理消除复杂背景因素对植物识别的负面影响.提出一种基于形态学变换的标记分水岭算法,对植物叶片进行重建的开闭操作,并使用标记分水岭算法对其进行分割.识别测试纹理与形状特征提取方式,了解图像分割算法对复杂背景消除的有效性.结果表明:提出的算法能有效分割复杂背景叶片.

Research of Plant Recognition Using Segmentation and Feature Fusion

Considering of the existence of complex background, the negative effects of complex background factors on leaves recognition are eliminated by image segmentation. At the same time, a segmentation algorithm be founded on morphological transform is proposed to solve the over-segmentation problem in the segmentation of plant leaves. After the opening and closing based on the reconstruction, the plant leaves are segment by marker-watershed algorithm. The texture and shape features, extracted form the segmented leaves, are used to recognize the leaves. And the effectiveness of the recognition validates that the segmentation algorithm proposed in this paper is effective.