基于区域分割的图象压缩算法研究与应用

【作者】 元燕辉；

【导师】 杨长生；

【作者基本信息】 浙江大学 ， 计算机系统结构， 2002， 硕士

【摘要】 本文通过对图象区域分割理论的研究，提出一种基于区域分割的图象压缩算法，并给出该算法在基于图象内容检索方面的应用。该图象压缩算法主要是通过利用图象的分割对图象空间结构的认识，提高图象空间预测器的预测效率，从而达到提高图象压缩效率的目的。另外，对图象进行分割不仅仅给图象压缩带来了好处，也为进行基于内容的图象检索打下了基础。本文充分利用这一点，将图象压缩和内容检索两者统一起来，使得图象的区域化信息达到最大化的利用，从而提高单位数据的信息含量。 在本文中，图象的区域分割不仅是图象压缩算法的基础，也是图象内容检索应用得以实现的基础，本文第二章将专门对图象区域分割的各种算法进行介绍。利用适当的图象区域分割算法，可以总结出图象的空间分布规律，从而能够采用更方便、更有效的图象压缩算法对图象进行压缩。采用这种方式对图象进行压缩以后，图象可以分成一系列的图象区域块，每个区域块具有自己的压缩特征参数，这些特征参数可以作为图象检索的关键字，这大大提高了图象内容检索的效率。另外，图象进行区域化压缩以后，用户可以选择解压哪个特定区域，这样用户可以通过观察区域的情况来判断整个图象是否符合查询要求，而不必传输整个图象数据。本文在第三章中将给出基于区域分割的图象压缩算法模型，并给出该模型需要注意的一些问题；在第四章给出图象进行区域化压缩以后应该如何进行组织，从而实现图象的内容检索及提高检索的效率；第五章给出一些图象区域分割和图象压缩的试验结果，另外，对图象内容检索进行预测性分析。 本文的算法的思想比较新颖，综合了多个图象领域的要素来考虑问题，问题的思考模式不是单纯的问题—解决方法模式，而是问题—解决方法—面向解决方法的应用，如此将加深对问题解决方法的选择考虑，使得算法更加有效，更加具有实际意义。更多还原

【Abstract】 Based on researching of image segmentation theoiy, an efficient image compress algorithm is presented in this paper. Moreover one important application about this algorithm is discussed, and this method is often used in image index based on context. More spatial structure information in images, the image spatial predictors are more efficient. Image segmentation theory just gives a chance to achieve the spatial structure information of images. In addition, this compress algorithm based on image segmentation brings some advantages in image index. In this paper, image compress algorithm and image index based on context are unified by image segmentation theory, and this would maximize the utility of information data. Image segmentation plays a very important role in the algorithm. Not only image compress algorithm is based on it, but also image index based on context is realized through it. Image can be segmented to some regions after compressed by the algorithm based on image segmentation. And each region has its compress parameters, which could be used as key words for image index. As a result, the efficient of image index is improved. Moreover, a user could choose some region to be decompressed if this image is compressed by the algorithm, and he could also examine some region to decide whether this image is required instead of examine the whole image. When this algorithm is used in a net for image index, the image data needed to be transferred could be reduced. This algorithm has synthesized a few important factors in the field of image, and it introduces a very novelty method.更多还原