Design and Development of an Image Based Plant Identification System Using Leaf

Because of huge diversity of plants existing on earth and large inter-species similarity, the manual process of identification of plants is difficult and at times, the results generated may be confusing. So, it is necessary to automate the process of identification of plants to generate faster and accurate results. Among various parts of plant, leaf is easily available in all seasons and can be easily scanned for understanding the shape details. So, in this paper, we have worked on plant identification using leaf image. In the process of identification of plants, the key challenge identified is to keep the size of feature vector reasonable and still achieve accurate and effective results. As the results are to be interpreted by the user, it was necessary to understand the theories behind human perception to reduce the semantic gap. We have designed and developed a tool ‘SbLRS: Shape based Leaf Recognition System’ which is a three-staged system that performs identification of plant using leaf shape. Our system can be helpful for following userscommon layman person, as replacement for biologists in remote locations, as teaching aid for teachers and students, for a farmer to identify the plant species and to assist biologists in identification process. The parameters important for identifying leaf image according to human perception are identified and defined at three different levels. On the basis of results generated, the effectiveness of SbLRS is compared with existing contour based methods in terms of recall, precision. It was observed that our tool SbLRS showed satisfactory results for identification of plant using leaf at small feature vector size and simpler computations.