计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (9): 17-25.DOI: 10.3778/j.issn.1002-8331.1612-0150

• 热点与综述 • 上一篇    下一篇

基于形状特征的叶片图像识别算法比较研究

陈良宵1,王  斌1,2   

  1. 1.南京财经大学 信息工程学院,南京 210023
    2.南京财经大学 电子商务省级重点实验室,南京 210023
  • 出版日期:2017-05-01 发布日期:2017-05-15

Comparative study of leaf image recognition algorithm based on shape feature

CHEN Liangxiao1, WANG Bin1,2   

  1. 1.School of Information Engineering, Nanjing University of Finance and Economics, Nanjing 210023, China
    2.Key Laboratory of Electronic Business, Nanjing University of Finance and Economics, Nanjing 210023, China
  • Online:2017-05-01 Published:2017-05-15

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。

关键词: 植物分类, 叶片识别, 图像处理, 形状分析, 形状识别

Abstract: Plant is one of the main forms of life, and the number of the known plant species has reached about 400,000. The classification and identification of plants play an important role in biodiversity conservation, ecological agriculture and biological safety. Different species of plants usually have different leaf shapes, so the shape features provide key cue for plant classification. Leaf image recognition, as a significant application of computer vision, has received considerable attentions and has been made great progress in recent years. However, due to the issues of the huge amount of plant species, the large intra-class variance, the small inter-class differences and the leaf self-intersection, the leaf image recognition is still an unsolved problem. This paper presents a review over the existing shape based algorithms for leaf image recognition. In this review, the existing shape based leaf recognition algorithms are classified and several state-of-the-art algorithms of them are further analyzed and compared. The widely used leaf image databases and performance evaluation methods are also introduced. An experimental study on several types of representative shape based algorithms for leaf image recognition is also conducted. This paper not only presents a guide of applying the existing shape based algorithms for leaf recognition, but also gives a direction for developing novel algorithms of higher performance for leaf recognition.

Key words: plant classification, leaf recognition, image processing, shape analysis, shape recognition