Cascade-Forward Neural Network in Identification of Plant Species of Desert Based on Wild Flowers

Tremendous improvements in Flower image description induced much interest in image based plant species identification. Rare species of desert plants are at risk and it is necessary to maintain record for their existence, which can be done by applying image processing techniques for object classification. This paper focuses on the automatic recognition of plant species from Sonoran desert regions through their flower images. The dataset contains 609 individuals of 25 species. The image preprocessing begins with median filter to remove the noise. The color and texture features are obtained from the flower images for classification. HSV color space is used to extract the color features and Center-Symmetric Local Binary Pattern (CS-LBP) for texture features. The extracted features are incorporated in Cascade-Forward Neural Network to classify the species which outperforms an accuracy of 96.8%.