Research On Nude Picture Algorithm Base On ICA Algorithm And SVM Algorithm

As the development of internet and the network technology getting more and more popular, internet gradually became the sharing and communication tool for people. But the development of network is also has its defects, one of which is the spread of nude picture, blocking the sending and the spread of nude picture seem so important now. Before we can block the spread of the nude pictures we should distinct the nude pictures from the normal pictures first.Similar to other pattern recognition problem, nude picture detection problem can be divided into features extraction and classifier training. And the features of nude pictures can be divided into features recognized by human and features recognized by computer. PCA(principal component analysis) is one of the most popular and vital algorithms in processing images. But PCA can only remove the linear correlation of the data. ICA is a signal processing technology develop from blind source separation problem, comparing to PCA, it make the elements separated independent from each other, not just nonlinear correlate with each other. ICA algorithm has become more and more popular these years.Artificial neural network and the bayes algorithm are two of the most popular algorithms of features classification, but also have their own defects. SVM(support vector machine) is invented by Vapnik, a machine learning technology base on statistical learning theory, it solve the problem that artificial neural network that can't solve ,SVM has applied at many field as an useful classification algorithm.