Multi-layer classifier and Internet image aided training-based color image emotion classification method

Abstract

The invention relates to a multi-layer classifier and Internet image aided training-based color image emotion classification method. The method comprises the following steps that: six kinds of color and texture features of color images are extracted to represent emotions contained in the images; a supervised hierarchical classification model is established; at the bottom layer of the model, a support vector machine (SVM) and an AdaBoost method are adopted to train a weak classifier for each kind of image features, and at the upper layer of the model, the AdaBoost method is adopted again to combine six weak classifiers corresponding to the six kinds of features, so that a strong classifier can be formed; as for each training image, a certain number of images similar to the corresponding training image are retrieved from the Internet to form a larger training image set; and finally, the larger training image set and the original training images are adopted together to train the hierarchical classification model through a supervised learning mode, and an emotion-based color image classifier can be obtained.

Description

technical field [0001] The present invention relates to an image classification method, in particular to an image classification method that uses images retrieved from the Internet for auxiliary training, and combines a Support Vector Machine (Support Vector Machine, SVM) and AdaBoost in multiple layers based on the transmitted emotion. Color Image Classification Methods. Background technique [0002] The technical research of color image classification based on the emotion conveyed by images mainly involves three key issues: the establishment of image emotion model, the extraction of image emotion features, and image emotion recognition. The emotion model of image can be divided into discrete model and continuous model. In the continuous model, emotions take different continuous values ​​in several dimensions. For example, a commonly used continuous model is the dimensional observation model proposed by Mehrabian and Russell in 1974—the PAD three-dimensional emotional mode...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the problem of low correct rate of emotion classification methods of existing color images, the present invention provides a color image based on multi-layer classifier and Internet image auxiliary training. Image Sentiment Classification Method

Images