被动毫米波图像超分辨复原算法及DSP实现

发布时间：2018-06-19 09:24

本文选题：被动毫米波成像 + 超分辨复原　；参考：《电子科技大学》2012年硕士论文

【摘要】：被动毫米波成像是通过接收物体毫米波频段的辐射能量差异来实现成像。与光学或X光成像技术相比,因其穿透性强、无辐射、隐蔽性高,在军事侦察、航空安检、反恐斗争等领域具有重要应用价值,也是国内外研究的热点。超分辨复原是被动毫米波成像信号处理的关键技术,能恢复由成像系统低通效应所丢失的带外的高频信息,提高成像分辨率。本文针对被动毫米波图像超分辨复原算法及DSP实现问题,主要研究内容如下： 1.分析对比Richardson-Lucy算法、投影Landweber算法、图像空间重构ISRA算法等基于统计优化理论的超分辨复原算法,给出了各算法的计算复杂度,为超分辨算法DSP实现奠定基础。 2.提出了基于PCA和小波分解的自适应正则化超分辨算法,能够有效利用毫米波图像局部灰度相关性,与传统超分辨算法相比,有更好复原效果。 3.基于ADSP TS-201硬件平台,完成了实时信号处理单元方案设计,实现了数据传输通信,定标通道校正、图像数据重排、通道均衡去条带等图像预处理及超分辨算法函数库等功能。 4.完成了被动毫米波成像系统的联调测试,以及针对多种目标和场景的单通道和多通道系统外场成像实验。最后通过仿真和实测数据验证了基于PCA和小波分解的自适应正则化超分辨复原算法的有效性和成像系统信号处理单元的功能性能。
[Abstract]:Passive millimeter wave imaging is realized by receiving radiation energy difference of millimeter wave band. Compared with optical or X-ray imaging technology, because of its strong penetration, no radiation, high concealment, it has important application value in military reconnaissance, aviation security inspection, anti-terrorism struggle and so on, and it is also a hot research topic at home and abroad. Super-resolution restoration is a key technique in passive millimeter-wave imaging signal processing. It can recover the out-of-band high-frequency information lost by the low-pass effect of the imaging system and improve the imaging resolution. In this paper, the passive millimeter wave image super-resolution restoration algorithm and DSP implementation problem, the main research contents are as follows: 1. Compared with Richard son-Lucy algorithm, projection Landweber algorithm, image space reconstruction ISRA algorithm and so on, the algorithm of super-resolution restoration based on statistical optimization theory is analyzed and compared. The computational complexity of each algorithm is given, which lays a foundation for the realization of super-resolution algorithm based on DSP. 2. An adaptive regularized super-resolution algorithm based on PCA and wavelet decomposition is proposed, which can effectively utilize the local gray correlation of millimeter-wave images. Compared with the traditional super-resolution algorithm, the algorithm has better restoration effect. Based on ADSP TS-201 hardware platform, the scheme design of real-time signal processing unit is completed, and data transmission communication, calibration channel correction, image data rearrangement are realized. Channel equalization, strip removal and other image preprocessing and super resolution algorithm functions. 4. The test of passive millimeter-wave imaging system is completed, and the field imaging experiments of single-channel and multi-channel systems for multiple targets and scenes are also carried out. Finally, the validity of the adaptive regularized super-resolution restoration algorithm based on PCA and wavelet decomposition and the functional performance of the signal processing unit of the imaging system are verified by simulation and measured data.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.41;TP368.1

【参考文献】