摘要
为增强现有的风格化效果,提出一种基于直方图匹配和图像修复的图像风格化处理方法。根据超像素分割算法对源图像和参考图像进行分块,将每块进行标记;根据定义的最佳匹配算法,即以块为单位的平均值、方差、色调及平滑度对比,找到源图像在参考图像中的最佳匹配块;通过直方图匹配,将参考图像的风格采用最佳匹配算法原则以块为单位转换到源图像中去;根据图像修复算法修复块与块的拼接处,减少输出图像的噪声。实验结果表明,该方法可以适应不同的风格化要求,保留图像本身内容,效果很好。
To enhance the existing stylized effects,a method of image stylization based on histogram matching and image inpain-ting was presented.The source image and the reference image were found were divided into blocks according to the superpixel segmentation algorithm,and each block was marked.The best matching block of the source image in the reference image were found according to the best matching algorithm which were the average value,the standard deviation,the hue and the smoothness compare in blocks.By using histogram matching algorithm,the style of the reference image was transferred to the source image according to the best matching algorithm in blocks.Image inpainting algorithm was used to repair the splicing of the block to reduce the noise of output image.Results of the simulation indicate that the proposed method can be adapted to different stylized requirements,which preserves the contents of the images themselves and works well.
作者
叶红娟
彭宏京
杜振龙
YE Hong-juan;PENG Hong-jing;DU Zhen-long(College of Computer Science and Technology,Nanjing Tech University,Nanjing 211800,China)
出处
《计算机工程与设计》
北大核心
2018年第10期3176-3180,3200,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(61672279)
关键词
基于实例的风格化
图像分割
最佳匹配
直方图匹配
图像修复
example-based stylization
image segmentation
best match
histogram matching
image inpainting
