Mobile-based flower species recognition in the natural environment

Abstract

A system design and new method for recognising a flower's species in a natural environment are proposed. An experimental analysis is conducted to improve the optimal method of feature extraction that is developed for the colour, texture and shape. These results showed that the colour (HS, CbCrCg and RGB), texture (grey level co-occurrence matrix) and shape (zero crossing rate) features for pictures of flowers were the most effective. The study develops flower image recognition technology that can be utilised in a mobile environment.