Crop disease detection and classification based on hybrid information approach

The objective of this paper to identify the diseases in the leaves of the all plants. Plant disease diagnosis helps to improve both the quality and quantity of crop productivity. In existing, to detect the diseases they used the spectroscopic techniques. These techniques are very expensive and can only be utilized by trained persons only. This work proposes an approach for the detection of leaf diseases based on the characterization of texture, shape and color properties. The detection of diseases which are detected using ISRC(improved sparse Representation Classifier) technique. First the GENABC clustering approach is applied to the input image to segment the affected area. Then extract the features from the affected area by using feature extraction techniques. In this paper Improved Transform Encoded Local Pattern used to extract the texture feature, Enhanced Gradient Feature (EGF) to extract the shape and Improved Color Histogram Techniques(ICH) are used to extract the color. And then these features are given to the ISRC classifier to get the exact type of disease on affected leaves. To analyze the performance of the proposed method we use four metrics. They are classification accuracy, error rate, precision value and recall value. From the analysis of experimental results, the ISRC method provides the best result than the existing approach.