CAJ | 학술논문

多媒体技术的发展导致数字图像迅速增长，如何根据语义特征高效检索出满足用户要求的图像，已成为当前各行业迫切需要解决的问题。为此提出一种基于颜色、纹理和形状三种语义特征的图像检索方法，建立了颜色和纹理特征的语义描述，使用BP神经网络实现了低层视觉特征到高层语义特征的映射。选取Corel图像库作为测试图像库，实验通过与基于颜色语义特征的检索方法相比较，取得了良好的实验效果。
다매체기술적발전도치수자도상신속증장，여하근거어의특정고효검색출만족용호요구적도상，이성위당전각행업박절수요해결적문제。위차제출일충기우안색、문리화형상삼충어의특정적도상검색방법，건립료안색화문리특정적어의묘술，사용BP신경망락실현료저층시각특정도고층어의특정적영사。선취Corel도상고작위측시도상고，실험통과여기우안색어의특정적검색방법상비교，취득료량호적실험효과。
The development of multimedia technology has resulted in the rapid growth of digital images. It has become an urgent problem to retrieve images which can better meet the users' requirements according to semantic features. A method on image retrieval based on three local features of color,texture and shape is proposed. The method builds semantic description of color and texture features,uses BP neural network to implement the mapping between low visual features and semantic features. Using Corel image database as testing image database,experiments have achieved good effect compared with the method only based on color semantic feature.