48 / 2015-10-28 11:03:03

Sunflower diseases recognition algorithm based on wavelet domain feature dimension reduction

Wavelet analysis; Feature Vector Dimension Reduction; Probabilistic neural network

A new sunflower diseases recognition algorithm has been presented, which is based on image processing and pattern recognition. Firstly, sunflower diseases were collected for segmenting disease spot. The color histogram based on RGB space and color features based on HSI space of leaf for disease are extracted, based on gray level co-occurrence matrix texture features, these features are arranged for one dimensional vector. Dimension of characteristic vector is reduced by wavelet analysis, and the original vector is replaced by discrete approximation information of characteristic vector. Finally, the dimension reduction series and identify diseases were trained and automatically determined by probabilistic neural network.
Experimental results show that the system not only can accurately identify these three diseases, sunflower powdery mildew, sunflower black rot and sunflower downy mildew, but also can make the feature vector have low characteristics dimension, in the mean time, ensure the recognition accuracy.