结合形状先验的图像分割算法及应用研究

发布时间：2018-01-29 10:51

本文关键词： 自动图像分割前景关键点局部形状匹配星形形状先验剪影风格化　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：图像分割是计算机视觉、模式识别和数字图像处理领域的热点研究问题。图像分割可分为交互式分割和自动分割两大类别。近年来,交互式分割虽然取得了很大的进展,但是处理大量图片时存在操作复杂、耗时耗力、对交互区域敏感等问题,因此自动分割方法受到研究者的广泛关注。为此,本文设计了一种自动图像分割算法,通过显著性检测算法和局部形状匹配算法自动检测前景的关键点,并将其应用到星形先验分割算法中,从而实现了图像的自动分割。本文主要内容和创新点如下:(1)前景关键点的提取对图像自动分割起决定性作用。为了自动标记图像前景,提出了一种基于局部形状匹配的前景关键点提取方法。该方法采用局部形状匹配来自动标记前景,并与基于显著性检测的前景标记方法相结合,能检测出更多的前景关键点,有利于后续的自动分割。(2)分割时结合形状先验知识能提高图像分割的准确性。星形形状是一种普适的先验知识,适用于自然界中的很多物体。本文实现了一种结合多星形形状先验的Graph Cuts分割算法,有效地提高了分割质量。(3)将提出的图像自动分割算法应用于图像的特效处理中,提出了一种基于样例的图像剪影风格化算法。该算法在图像分割的基础上,对前景和背景采取不同的色调处理策略,使生成的图像具有与样例图像类似的风格。实验结果验证了方法的有效性。
[Abstract]:Image segmentation is a hot research problem in the field of computer vision, pattern recognition and digital image processing. Image segmentation can be divided into two categories: interactive segmentation and automatic segmentation. Although great progress has been made in interactive segmentation, there are some problems such as complicated operation, time-consuming and labor-consuming, sensitive to interactive region, etc., so automatic segmentation method has been widely concerned by researchers. In this paper, an automatic image segmentation algorithm is designed, which detects the key points of the foreground automatically by the salience detection algorithm and the local shape matching algorithm, and applies it to the star prior segmentation algorithm. The main content and innovation of this paper are as follows: 1) extraction of foreground key points plays a decisive role in automatic image segmentation. In order to automatically mark the image foreground. This paper presents a method of extracting the key points of foreground based on local shape matching, which uses local shape matching to mark the foreground automatically, and combines with the foreground marking method based on salience detection. It can detect more foreground key points and can improve the accuracy of image segmentation by combining shape priori knowledge with subsequent automatic segmentation. Star shape is a kind of universal prior knowledge. It is suitable for many objects in nature. In this paper, a Graph Cuts segmentation algorithm combining multi-star shape priori is implemented. The proposed automatic image segmentation algorithm is applied to the special effect processing of the image. An image silhouette stylization algorithm based on sample is proposed. Based on image segmentation, different hue processing strategies are adopted for foreground and background. The generated image has a similar style to the sample image. The experimental results show that the method is effective.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】