基于HSV空间改进的多尺度显著性检测
发布时间：2019-06-06 03:35
【摘要】：图像显著性特征已被广泛地应用于图像分割、图像检索和图像压缩等领域,针对传统算法耗时较长,易受噪声影响等问题,提出了一种基于HSV色彩空间改进的多尺度显著性检测方法。该方法选择HSV色彩空间的色调、饱和度和亮度作为视觉特征,先通过高斯金字塔分解获得三种尺度的图像序列,然后使用改进的SR算法从三种尺度的图像序列中提出每个特征图,最后将这些特征图进行点对点的平方融合和线性融合。与其它算法的对比实验表明,该方法具有较好的检测效果和鲁棒性,能够较快速地检测出图像的显著性区域,能够突显整个显著性目标。
[Abstract]:Image saliency features have been widely used in image segmentation, image retrieval and image compression. In order to solve the problems of traditional algorithms, such as long time consuming and easy to be affected by noise, image salient features have been widely used in image segmentation, image retrieval and image compression. A multi-scale significance detection method based on HSV color space improvement is proposed. In this method, the hue, saturation and brightness of HSV color space are selected as visual features. Firstly, the image sequences of three scales are obtained by Gao Si pyramid decomposition. Then the improved SR algorithm is used to propose each feature graph from the image sequences of three scales. Finally, these feature graphs are point-to-point square fusion and linear fusion. Compared with other algorithms, this method has better detection effect and robustness, can detect the significant region of the image quickly, and can highlight the whole salient target.
【作者单位】： 淮阴工学院计算机工程学院;
【基金】：国家自然科学基金(61402192) 江苏高校自然科学研究计划(14KJB520006) 江苏省淮安市科技支撑计划(HAG2013068)
【分类号】：TP391.41
[Abstract]:Image saliency features have been widely used in image segmentation, image retrieval and image compression. In order to solve the problems of traditional algorithms, such as long time consuming and easy to be affected by noise, image salient features have been widely used in image segmentation, image retrieval and image compression. A multi-scale significance detection method based on HSV color space improvement is proposed. In this method, the hue, saturation and brightness of HSV color space are selected as visual features. Firstly, the image sequences of three scales are obtained by Gao Si pyramid decomposition. Then the improved SR algorithm is used to propose each feature graph from the image sequences of three scales. Finally, these feature graphs are point-to-point square fusion and linear fusion. Compared with other algorithms, this method has better detection effect and robustness, can detect the significant region of the image quickly, and can highlight the whole salient target.
【作者单位】： 淮阴工学院计算机工程学院;
【基金】：国家自然科学基金(61402192) 江苏高校自然科学研究计划(14KJB520006) 江苏省淮安市科技支撑计划(HAG2013068)
【分类号】：TP391.41
【参考文献】
相关期刊论文 前9条
1 张辰;杨文柱;刘召海;;基于HSV综合显著性的彩色图像分割方法[J];计算机工程与设计;2013年11期
2 卢凯;熊振海;李崇飞;李根;;基于全局结构相似度度量方法的显著性检测[J];计算机工程与科学;2013年06期
3 刘毅;黄兵;孙怀江;夏德深;;利用视觉显著性与图割的图像分割算法[J];计算机辅助设计与图形学学报;2013年03期
4 韦伟;张功萱;;基于HVS的多尺度显著性检测算法[J];计算机应用研究;2013年02期
5 于明;王倩;郭迎春;;一种图像的显著区域提取方法[J];光电工程;2012年08期
6 赵倩;曹家麟;胡越黎;;结合高斯多尺度变换和颜色复杂度计算的显著区域检测[J];仪器仪表学报;2012年02期
7 张巧荣;景丽;肖会敏;刘海波;;利用视觉显著性的图像分割方法[J];中国图象图形学报;2011年05期
8 邹国平;马儒宁;丁军娣;钟宝江;;基于显著性加权颜色和纹理的图像检索[J];山东大学学报(理学版);2010年07期
9 郁志鸣;王朔中;张新鹏;刘婷婷;;结合对比特性与局部清晰特性的图像显著区域检测[J];应用科学学报;2010年01期
【共引文献】
相关期刊论文 前10条
1 王文豪;周静波;高尚兵;严云洋;;基于HSV空间改进的多尺度显著性检测[J];计算机工程与科学;2017年02期
2 王健荣;王Y，
本文编号：2494034
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2494034.html
