Research And Implementation Of The Identification And Classification Of The Main Parts Of Plants Image

The identification and classification of plant images is a highly specializedjob.However, with the popularity of camera equipment, plant lovers and researchers hadcollected a large number of plant images. To identify and classify the massive plantimages brought tremendous pressure to the limited staff of plant classification.Computer-aided plant image identify and classify will greatly enhance the objectivity andefficiency of the classification.At the same time, helping professionals research extendsto a wider area.Digital image segmentation and main parts extraction is a highly concerned researchbrunch in the field of computer vision and image processing.Main part segmentation isan important step for image processing and image information extraction,and is also thebasis of image retrieval.Only after a reasonable image segmentation,can we get furtherinformation from the image.Cause of that most plants image are taken outdoor,there ismore or less a complex or simple background.It requires a use of computer imageprocessing techniques to exteract the main parts of a image for different images.Then weget the characteristics of the main parts of the plants image and classification. This papermainly uses503classes,16297plant images provided by the CAS Institute of Botany asan image database to apply our research on image segmentation algorithms andapplications. After the image segmentation, we carried out the identification of the mainparts of the plant image, extract the color features of plant images, and use support vectormachine (SVM) combined with classical clustering algorithm Kmeans to classify theplant images. At last, we established a classification system to identify the main plant andclassify the plant images.After the identification and classification of the16297plant images,recognition rateof the main parts of flower reached84.6%and leaf reached73.4%. Sorted under linearclassification success rate is86.5%, Gaussian classification Sorted success rate is88.25%.