基于形状特征的叶片图像识别算法比较研究

发布时间：2019-03-07 08:26

【摘要】：植物是生命的主要形态之一,其种类已达40多万种,对其进行分类识别在生物多样性保护,生态农业,生物安全中有着重要的意义。不同的种类的植物一般有着不同的叶片形状,因此叶片的形状特征在植物分类中扮演着重要的角色。作为计算机视觉的一个重要应用的植物叶片图像识别,近些年来受到了学者们的关注,产生了大量的研究成果。但由于植物种类巨大,叶片图像存在的类内差异大、类间差异小和叶片的自遮挡等问题等诸多问题,使得叶片图像的识别仍然是目前计算机视觉应用研究的一个热点。对近些年来的基于形状特征的叶片图像识别算法进行了综述和比较,对现有的算法进行了分类,对目前各类最先进的识别算法进行了分析和比较。此外,还介绍了常用的叶片图像测试集和性能评估方法,并将各类算法进行了实验结果的比较研究。研究工作既为现有的植物叶片识别算法的实际应用提供了指导,又为今后进一步研究新的高性能的识别算法提出了努力的方向。
[Abstract]:Plant is one of the main forms of life, and its species have reached more than 400,000 species. It is of great significance to classify and identify plants in biodiversity conservation, ecological agriculture and biosafety. Different species of plants usually have different leaf shapes, so the shape characteristics of leaves play an important role in plant classification. As an important application of computer vision, plant leaf image recognition has been concerned by scholars in recent years, resulting in a large number of research results. However, due to the huge plant species, large intra-class differences in leaf images, small inter-class differences and leaf self-occlusion and other problems, leaf image recognition is still a hot topic in the application of computer vision. This paper summarizes and compares the blade image recognition algorithms based on shape features in recent years, classifies the existing algorithms, and analyzes and compares the most advanced recognition algorithms at present. In addition, the commonly used blade image test sets and performance evaluation methods are also introduced, and the experimental results are compared with each other. The research work not only provides guidance for the practical application of existing plant leaf recognition algorithms, but also puts forward the direction of further research on new high-performance recognition algorithms in the future.
【作者单位】：南京财经大学信息工程学院;南京财经大学电子商务省级重点实验室;
【基金】：国家自然科学基金(No.61372158) 江苏省2015年普通高校研究生科研创新项目(No.KYZZ15_0277)
【分类号】：TP391.41

【相似文献】