Effects of Indices on Temporal Data for Specific Crop Identification by Using Possibilistic-C Means

Remotely sensed images of the Earth is useful source of information for monitoring agricultural crop production. These images are available over a large range of temporal and spatial scales. Since objects (including vegetation) have their unique spectral features (reflectance or emission response), they can be identified from remote sensing imagery according to their unique spatial characteristics. The strong contrast of absorption and scattering of the red and near infrared bands can be combined into different quantitative indices of vegetation conditions Four spectral indices SR (Simple Ratio), NDVI (Normalized Difference Vegetation index), TNDVI (Transformed Normalized Difference Vegetation Index) and SAVI (Soil-Adjusted Vegetation Index) using conventional and CBSI(Class Based Sensor Independent) were investigated to identify a single crop using temporal multi-spectral images. To extract a single crop from coarse resolution data encounters problem of ‘mixed pixel’ (i.e. multiple classes within a single pixel). The mixed pixel problem can be handled using soft computing techniques. Fuzzy classification approach is one of soft computing technique can be beneficial where a mixed pixel may be assigned multiple class membership. In this experiment CBSI and conventional indices data has been taken as input in Possibilistic c-means classifier, and cotton crop has been identified at sub-pixel level. The classification results with respect to the conventional and CBSI indices are compared in terms of image to image maximum classification accuracy and found overall accuracy of 97.14 % for NDVI of Dataset 2. Data used for this study was AWIFS for soft classification and LISS-III data for soft testing data generation from Resourcesat-1(IRS-P6) satellite. The research indicates that appropriately used vegetation indices can incorporate temporal variations while extracting specific crop of interest with soft computing techniques for coarser spatial resolution temporal remote sensing data.