image Land cover refers to the surface cover of land, be it vegetation, urban infrastructure, water, bare soil or otherwise. Identifying, delineating and mapping land cover is important for global monitoring studies, resource management and planning activities. Crop monitoring information is very important for food security and helps improve our knowledge on the role of agriculture on climate change and crop type identification. This work focuses on an automated KNN classification system to identify land cover and crop type in synthetic aperture radar (SAR) images. In the first module, an unsupervised Kohonen Self-Organizing Mapping (SOM) neural network is used to identify the terrain type. Say no to plagiarism. Get a tailor-made essay on "Why Violent Video Games Shouldn't Be Banned"? Get an original essay In the second module, local binary pattern (LBP) based features are extracted to identify the crop type in the area covered by the crop. The extracted features are entrusted to the KNN classifier which classifies the type of crop. Introduction Agriculture is the backbone of the Indian economy, where around 70% of the population depends on agriculture. In agriculture, parameters such as canopy, yield and product quality were important measures from the farmers' point of view (Viraj et al,2012). India is the major producing country of many crops. The major crops in India can be divided into four categories, viz. Food grains, cash crops, plantation crops and horticultural crops. Learning Multistage and Deep Representations to Classify Remotely Sensed Images (Zhao et al 2016) Land cover refers to the surface cover of land, be it vegetation, urban infrastructure, water, bare soil, or other. Identifying, delineating and mapping land cover is important for global monitoring studies, resource management and planning activities. Land cover identification establishes the baseline from which monitoring (change detection) activities can be performed and provides land cover information for thematic baseline maps. Crop monitoring information is very important for food security and helps improve our knowledge on the role of agriculture on climate change, crop type identification, land cover, etc. (Ajay et al 2012) Measuring crop types leads to numerical descriptions of the crop, helps determine a problem that is big enough to solve or small enough to ignore. CNN-based 3-D FE model with combined regularization to extract effective spectral-spatial features of hyperspectral images. The proposed 3-D deep CNN provides excellent classification performance under limited training samples. Designing adequate deep CNN models is quite difficult. Nataliia Kussul1 et al(2016) proposed the methodology to solve large-scale area classification and estimation problems in remote sensing domain based on deep learning paradigm. It is based on a hierarchical model that includes self-organizing maps (SOM) for data pre-processing and segmentation (clustering), multi-layer perception set (MLP) for data classification and fusion of heterogeneous data and geospatial analysis for post-processing. Shoulda set of methods can be exploited (“expert mix” approach) to take advantage of different processing methods and techniques. Kernel function processing in clustering takes more time. Christopher McCool et al (2016) proposed a new crop detection system applied to the challenging task of field pepper detection. Cropping field-grown sweet peppers presents several challenges for robotic systems, such as the high degree of occlusion and the fact that the crop can have a similar color to the background (green on green). To overcome these problems, they proposed a two-stage system that performs per-pixel segmentation followed by region detection. This approach has the advantage of providing robustness against occlusion (since features are only taken from a small region) and minimizing the amount of laborious annotation (since only the crop class needs to be annotated). The accuracy of crop segmentation is low. Adriana Romero et al (2016) proposed unsupervised pre-training of the Greedy layer coupled with a highly efficient algorithm for unsupervised learning of sparse features. The algorithm is rooted on sparse representation and simultaneously enforces both population and sparsity over time of the extracted features. The advantage of using spatial information is that the combination of a large number of output features and max-pooling steps in deep architectures are crucial to achieve excellent results. To access the generalization of encoded features in multi-temporal and multi-year image settings J. Théau et at (2016) describes that overview of change detection techniques applied to Earth observation and used methodologies such as image differencing, of principal components, post-classification comparison, change detection technology. The main takeaway from the paper is that change detection algorithms have their own merits and no single approach is optimal and applicable to all cases. Data selection is a critical step in change detection. Summary Traditional unsupervised classification algorithms, such as maximum likelihood classification, use clustering techniques to identify spectrally distinct groups of data and represent the first automatic land cover classification approach using pattern recognition techniques . The disadvantage of these algorithms is that the accuracy of land cover classification is not guaranteed and the land cover classifications are arbitrary. Supervised classification methods require substantial expertise and human participation for the selection of training samples. Therefore, the result of land cover classification is strongly influenced by the classification participants, and it is impossible to automatically classify land cover with these methods. Furthermore, algorithms such as neural network classification and fuzzy logic classification are very complicated in their algorithmic basis, making them difficult to understand and apply on a large scale. Decision tree classification methods are widely used in large areas, such as global land cover mapping. The main problem presented by decision tree classification is the construction of the decision tree and the assignment of thresholds for each subnode, which is highly dependent on human experience and varies spatially and temporally. Proposed work. Proposed system architecture The proposed architecture..