Scene classification, the classification of images into semantic categories (e.g., coast, mountains, highways and streets) is a challenging and important problem nowadays. Many different approaches and feature extraction methodologies concerning scene classification have been proposed and applied in the last few years. In real time environments, we prefer a feature extraction method which helps us with minimal data, performing better with less execution time. In this aspect, we are proposing hybrid feature extraction methods (hybrid-1, hybrid-2 and hybrid-3) which includes geometrical, statistical and texture features for natural scene categorization problems. The results are proving the efficiency of the proposed (hybrid-3) feature extraction method over the commonly used feature extraction methods such as geometrical moments, statistical moments and texture features. These features are tested using radial basis kernel functions (n=5) in support vector machine. Images are used without any preprocessing, making the system robust to real scene environments. This complete work is experimented in Matlab 6.5 using real world dataset.