基于颜色、形状的情感图像检索

Semantic Image Retrieval Research Based on Color and Sharp

图像所含的内容丰富多彩,是用物理特征无法完整表达的。目前常用的图像检索技术是基于内容的图像检索,检索出的图像在某些程度上无法满足用户的需求。提出了一种基于颜色和形状的情感图像检索方法,讨论了图像的物理特征——颜色和形状的提取与表达。分析不同颜色、特征所激起人的不同情感的反应,建立情感库,用模糊神经网络建立底层物理特征到高层语义特征的联系,实现基于情感语义的图像检索,实现真正的以人为主导的信息检索,提高人机交互的和谐性和检索效率。

Image is used to convey emotion, if only use the image＇s physical features can＇t expression prefect, because the content of the image is abundant. At present, traditional image retrieval method is based on the content of the image ; the result can＇t be satisfied with people on some degree. A new method is proposed which base on the color and sharp feature, analysis the method which retract and express the color and sharp, analysis different color and sharp arouse people＇s different emotion, building emotion database, use fuzzy neural network combine the low level physical feature with the high level semantic feature, achieve the retrieval method which base on the emotional semantic, make the man stand at the center during index the information, improve the harmonious and efficiency between man and machine.