Visual Cultural Symbol Recognition Based on Muti-Feature Extracting

Visual cultural symbol (VCS) is common around us, and it´s very important for cultural study. Especially, it´s useful and efficient if we can study VCS through CS. In the Internet, Mostly VCS display as an image. Machine learning is emerging, but image recognition is focused on face or man detective, it´s lack of VCS recognition study. Content based image classification techniques are gaining increasing popularity in the visual contents of the images for classifying images to their categories of interest which has been accomplished using various techniques. Feature extraction is an important part of the classification process. In this paper, we mainly using HOG (Histogram of Oriented Gradient), LBP (Local Binary Pattern), RGB, to extract the VCS´s contour, texture, color features, respectively, and also comprehend them. The efficiency of feature extraction techniques to implement according to above methodologies are tested using the Support Vector Machines (SVM) Classifier. In all these techniques different degrees of image classification precision and recall has been calculated. There are some impressive conclusions which are made from our experiments.