Bp 407 en
Recognition of Plants Based on Images of Fruit
Author: Jan Kůrka
This thesis deals with semi-automatic plant identification based on photographs of fruit in the natural environment. The method proposed in this thesis has two stages: segmentation and recognition. For the segmentation stage, a semi-automatic method based on a combination of graph cut and color histogram is presented and evaluated. In the recognition stage, the foreground is processed by a nearest neighbor classifier. As features, the color histogram and axes ratio of an ellipse fitted by Direct Least Square Fitting method are used.
- Kůrka Jan, tel: +420 736 725 426 , mailto:email@example.com