随着科学技术的飞速发展，机器学习与人工智能技术的不断创新，人们对特定信息检索的需求逐渐增加，使得如何对资源进行合理有效的分类成为一个关键问题。近几年来，基于内容的图像分类的研究焦点主要集中在自然图像的场景分类和物体分类两个方面，大多采用有监督学习方法，通过对底层特征建模和中间语义分析来实现分类。
本文基于Libsvm的图像分类研究及实现，主要针对的是物体分类这一方面，选用了五类水果作为分类研究的对象。对图像进行分类的大体步骤主要包括采集图像样本（主要从Web上获取）、图像预处理（如截成大小一致的图片）、特征向量提取、结合Libsvm进行模型训练、对测试图片进行分类测试。(With the rapid development of science and technology, the continuous innovation of machine learning and artificial intelligence technology, the demand of people for specific information retrieval increases gradually, and how to classify resources reasonably and efficiently becomes a key issue.
This article based on Libsvm image classification research and implementation, mainly for the object classification on the one hand, the selection of five types of fruit as the classification of the object of study. The general steps of image classification include collecting image samples (mainly obtained from the Web), image preprocessing (such as cutting the same size of the image), feature vector extraction, combined with Libsvm model training, the test images were classified test.)