基于自适应阈值的均匀局部二值模式

Adaptive threshold-based uniform local binary patterns

针对均匀局部二值模式的编码方式存在不同类像素干扰的问题,提出一种基于自适应阈值的均匀局部二值模式。采用基于直觉模糊集理论相似度量的方法计算图像前景与背景的分割阈值,通过引入该阈值参数改进均匀局部二值模式的编码方式,使得改进后的编码方案能有效降低不同类像素间的干扰。将改进的均匀局部二值模式应用于现勘图像检索,实验结果表明,该模式可以在一定程度上降低不同类像素间的干扰,增强算法对现勘图像纹理特征的描述性能,且使得检索算法取得较好的检索结果。

To deal with the problem of interference from different classes of pixels in the coding mode of uniform local binary patterns,an adaptive threshold-based uniform local binary patterns is proposed.In this algorithm,the method of intuitionistic fuzzy set theory similarity measure is used to calculate the segmentation threshold of image foreground and background.By introducing the threshold parameter,the coding mode of the uniform local binary patterns is improved so that the improved coding scheme can effectively reduce the interference from different classes of pixels.The improved uniform local binary mode is applied to the crime scene investigation(CSI)image retrieval.Experimental results show that it can reduce the interference from different classes of pixels to a certain extent,enhance the algorithm to better describe the texture features of the CSI image,and enable the CSI image retrieval algorithm to achieve higher retrieval accuracy.