Flower image recognition

• Main Author: Tan, Sherene Qiu Hua
• Other Authors: Yu Yajun
• Format: Final Year Project
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/60834
• Institution: Nanyang Technological University

This thesis studies the techniques to recognize flowers from images. Three features of the flowers, the Colour, the number of petals and the distance between each petal are used as parameters to identify the flower. The techniques to identify flowers are classified into three sections, color-based, shape-based and distance-based. Color base techniques revolve around values of the RGB color channels performing image segmentation and extracting color values of the flower. AForge.NET library imaging filters are used under shape-based techniques. Distance-based techniques involve the Euclidean distance formula and loop algorithms in order to achieve the best flower search results. The search results are based on the combined normalized scores of each feature which improves the accuracy and eliminate the limitation of differences in flower image size available in the database. The accuracy of the search results improve when more features are extracted.