Research of water hazard detection based on color and texture features

In this paper, we focus on the need for water hazard detection based on the characteristics of the static water body in off-road environment, which includes three main sections: extraction of color and texture features, building SVM model and practical detection of water bodies. Based on the features of high intensity, low saturation and low texture of the water bodies existed in off-road environment. Saturation-value ratio feature extracted from hsv color space of water body region, combined with other four texture features conducted by gray level co-occurrence matrix constitute the five-feature vector. Training set is established from sample images after the images are well preprocessed. Then build the svm model based on the training set. Our task is to separate practical samples into two classes: water region and land region according to the predict result calculate by svm model. Experimental results demonstrate significant progress on detection of water body hazard in off-road environment, which effectively reduce the influence of illumination variation exert on detection when only using color feature to detect. © 2013 IFSA.