摘要
为了消除服装图像背景的影响,针对目前的GrabCut算法存在对图像局部像素值的变化敏感、时间开销大、边缘不准确等问题,提出了改进的GrabCut算法。在改进算法中,通过对梯度图像使用多尺度分水岭去噪增强了图像的边缘信息,减少了后续处理的计算量;通过采取熵惩罚因子最优能量函数减少了检索图像的有效信息丢失。将改进后的GrabCut算法引入基于内容的服装图像检索系统中,实验结果表明与同类方法相比,所提方法在检索显示准确性以及检索的平均查准率和查全率方面均有明显的提升。
In order to eliminate interference of clothing image background,GrabCut algorithm was introduced.But the current GrabCut algorithm is sensitive to local noise and time-consuming,and its segmentation edge is not accurate.To solve these problems,the multi-scale watershed algorithm to de-noise gradient image was employed,enhancing the image edge points and reducing the subsequent processing computation.To reduce the loss of image key features,we employed entropy penalty factor optimal segmentation energy function,reducing the effective information loss of image retrieval.And then we introduced the improved GrabCut algorithm to content based clothing image retrieval system.The experimental results show the method has obvious advance on the accuracy of retrieval effect than the existing algorithms.
出处
《计算机科学》
CSCD
北大核心
2016年第S2期242-246,共5页
Computer Science
基金
广东省自然科学基金(2016A030313717)
国家自然科学基金课题(61472135)资助
