基于CCSDS的遥感图像感兴趣区域压缩研究

发布时间：2018-02-13 13:08

本文关键词： 空间遥感图像压缩 CCSDS算法码流分配感兴趣区域 itti模型　出处：《中国科学院研究生院(长春光学精密机械与物理研究所)》2014年博士论文　论文类型：学位论文

【摘要】：随着空间遥感技术的发展，光学遥感图像分辨率越来越高，单位时间内获得的图像数据量越来越大。然而，空间遥感图像的传输和存储技术发展相对迟缓。因此，获得图像后，有必要对图像进行压缩编码处理。对于一幅图像而言，我们通常只关注其中的一部分区域或目标，即感兴趣区域，而其他区域称为背景区域。所以，在对图像进行压缩处理时，可以采用感兴趣区域压缩算法：对感兴趣区域进行无损压缩或低压缩比压缩，而对背景区域采用大压缩比压缩，从而既降低了图像传输对带宽的要求，又减少了感兴趣区域细节信息的丢失。论文主要研究一种基于CCSDS的空间遥感图像感兴趣区域压缩算法，并尝试采用基于视觉注意机制的itti模型来检测图像的感兴趣区域。论文以海洋监视卫星图像为研究对象，尝试采用itti模型来检测图像内的舰船目标等感兴趣区域。首先，研究了itti模型的算法处理过程：分析图像的多种特征，并将其融合生成特征显著图；然后采用胜者为王和返回抑制机制提取出视觉注意点；最后以该点为圆心，设置固定值为半径，划定圆形区域为显著区域。本文将视觉注意点的提取转移过程建立为电容阵列充电模型，并在算法中引入了离散矩变换，增强了图像纹理特征响应；由视觉注意点提取显著目标时，本文采用了阈值分割算法。实验结果表明，改进算法所提取的显著区域形状大小基本与目标一致，且显著区域包含背景少。与itti模型相比，改进算法更适合应用于海洋监视卫星图像舰船目标检测提取。本文探讨了SPIHT、JPEG2000以及CCSDS等图像压缩算法，并重点研究了CCSDS压缩标准。CCSDS将图像分为若干个不同的段，段与段之间独立编码，每段的纹理复杂度不同，所包含的信息量不同。本文采用梯度来衡量图像的纹理复杂度，，并据此提出了一种基于梯度的压缩码流控制算法，纹理越复杂的段，所分配的码流容量越大，纹理越简单的段，所分配的码流容量越小。实验结果表明，采用该码流控制算法以后，恢复图像的信噪比有所改进。本文根据CCSDS的压缩特点，提出了一种新的感兴趣区域压缩算法，将感兴趣区域和背景区域进行分割，分别作为两幅独立的图像进行压缩。在压缩前，首先将感兴趣区域掩膜编码，然后将码流按一定比例分配给感兴趣区域和背景区域。之后引入基于梯度的码流分配算法，依次对感兴趣区域和背景区域编码，从而实现基于CCSDS的感兴趣区域图像压缩。实验结果表明，该算法能够提高图像感兴趣区域的恢复效果。
[Abstract]:With the development of space remote sensing technology, the resolution of optical remote sensing image is getting higher and higher, and the amount of image data per unit time is getting larger and larger. However, the transmission and storage technology of space remote sensing image is relatively slow. It is necessary to compress and encode the image. For an image, we usually focus on only a part of the region or target, that is, the region of interest, while the other region is called the background region. So, when we compress the image, we usually focus on the region of interest. The region of interest compression algorithm can be used: lossless compression or low compression ratio compression for the region of interest, and large compression ratio compression for the background region, which not only reduces the bandwidth requirement of image transmission, but also reduces the bandwidth requirement of image transmission. This paper mainly studies a region of interest compression algorithm based on CCSDS, and tries to use the itti model based on visual attention mechanism to detect the region of interest. In this paper, the marine surveillance satellite image is taken as the research object, and the itti model is used to detect the region of interest such as the ship target in the image. Firstly, the algorithm processing process of the itti model is studied: analyzing the various features of the image. The feature salient map is generated by fusion, and then the visual attention point is extracted by the winner king and the return inhibition mechanism. Finally, the center of the point is set as the fixed value and the radius is set. In this paper, the extraction and transfer process of visual attention points is established as a capacitive array charging model, and the discrete moment transform is introduced into the algorithm to enhance the texture feature response of the image. In this paper, the threshold segmentation algorithm is used to extract salient objects from visual attention points. The experimental results show that the shape of significant regions extracted by the improved algorithm is basically the same as that of the target, and the significant regions contain less background. Compared with the itti model, The improved algorithm is more suitable for ship target detection and extraction from marine surveillance satellite images. In this paper, we discuss the image compression algorithms such as SPIHT JPEG2000 and CCSDS, and focus on the research of CCSDS compression standard .CCSDs divide the image into several different segments, and each segment has different texture complexity. In this paper, gradient is used to measure the texture complexity of the image, and a gradient-based compressed bitstream control algorithm is proposed. The more complex the texture, the larger the capacity of the allocated bitstream and the simpler the texture. Experimental results show that the SNR of the restored image can be improved by using the bitstream control algorithm. In this paper, according to the characteristics of CCSDS compression, a new region of interest compression algorithm is proposed. The region of interest and the background region are segmented as two independent images. Firstly, the region of interest is masked, then the code stream is allocated to the region of interest and the background region according to a certain proportion. Then, a gradient-based bitstream allocation algorithm is introduced to code the region of interest and the background region in turn. The experimental results show that the proposed algorithm can improve the restoration effect of the region of interest.
【学位授予单位】：中国科学院研究生院(长春光学精密机械与物理研究所)
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】