压缩基础上利用纹理进行图像检索的方法研究

STUDY ON IMAGE RETRIEVAL METHODS WITH TEXTURE ON THE BASIS OF COMPRESSION

一幅图像的各个相邻像素之间是相关的,在对一幅图像进行检索的时候,总是提取图像的整体特征,所以一幅图像包含的信息对于检索来说是有冗余的。如果可以把冗余的信息减少,那么就可以大大提高检索的速度。采用高斯金字塔对其进行有效压缩,在压缩版本下提取图像纹理特征进而进行图像检索,大大加快了检索的速度和效率。实验证明,这种方法是有效的。

Every adjacent pixel in an image is related to each other.When retrieving an image,it is always the feature of the entire image that is retrieved;therefore the information contained in an image is more or less redundant to us.If the redundant information can be reduced,the retrieval speed can be significantly promoted.The article employs Gaussian pyramid to perform efficient compression on it,then retrieves the image texture features with the compressed version to greatly accelerate the retrieval speed and efficiency.Experiments have validated the effectiveness of the method.