【作者】 杨彦从；

【导师】 王汝琳；

【作者基本信息】 中国矿业大学（北京） ， 控制理论与控制工程， 2010， 博士

【摘要】 将分形理论应用到视频监控图像的编码与处理中可以有效提升编码效率和改善识别方法。视频监控图像可以表示为以像素位置及对应颜色强度构成的三维空间曲面,其空间分布特点反映了图像纹理的特点,在大多数情况下具有分形特征。分形维数把图像的空间信息与颜色信息简单而又有机地结合起来,有效地体现了纹理的复杂程度。针对图像的不同区域,可以选择计盒维数CBD或布朗运动维数BMD定量描述这一图像子区的纹理特点。计算结果表明,图像中的纹理越复杂,分形维数也就越大,纹理越简单,分形维数也就越小,相似的纹理具有大致相同的分形维数。据此提出了基于布朗运动维数的块匹配算法,并结合区域划分实现了高效分形编码算法,通过对标准图像及具体视频图像的编码分析及效果比较,验证了该算法的有效性。在视频监控图像的处理中,提出了分维图的概念,即通过特定大小的滑动窗口扫掠图像表面,计算窗口区域的分形维数而得到的图像,借助分维图可以进一步对图像进行目标识别、边缘检测、图像增强等处理。更多还原

【Abstract】 By means of the application of fractal theory in the encoding and processing of video surveillance image, encoding velocity and recognition results could be improved greatly. A video surveillance image can be represented by a 3-D surface of the colour intensity of each pixel position. The spatial distribution of the surface reflects the feature of image texture and it usually appears to be a fractal. Fractal dimension show the complex extent of image texture by linking the space information and colour information of images. For different regions in an image, counting-box dimension (CBD) or Brownian-motion dimension (BMD) can be used to describe quantitatively the feature of a region texture. It shows that the more complex of the image texture, the larger the fractal dimension; otherwise the simpler of the image texture, the smaller the fractal dimension. In addition, the similar image textures have almost the same fractal dimensions. Accordingly a new block matching algorithm was suggested that is concerning to Brownian-motion dimension, and furthermore region segmentation was introduced in order to propose an efficient fractal encoding algorithm. The encoding result of some standard images and video images according to this efficient fractal encoding algorithm was discussed and it is proved that this algorithm is effectiveness. A conception of fractal dimension image was proposed for the process of video surveillance image, which is obtained by sliding a window with special size on the image and calculating the fractal dimension of the image in the sliding window. Fractal dimension image is helpful to further image process and analysis such as target recognition, edge detection, and image enhancement and so on.更多还原