High Resolution Terahertz Image Processing

Terahertz imaging is a kind of imaging which has strong application prospects inthe fields such as biomedical imaging, industry and security detection. However,relatively low definition, bad contrast, and low resolution have become the mainobstacles that hinder the application of terahertz imaging. Due to terahertz-wave’sspecial characteristics, terahertz imaging is confronted with its unique technicaldifficulties.In this thesis, efforts are dedicated to the investigation that combines both theimage processing techniques and terahertz imaging principles. By completing theprocesses of terahertz image de-noising and enhancement, suppressing the fluctuationeffects from terahertz radiation source and high resolution terahertz imagereconstruction, high quality terahertz images with low noise, good imagingtarget-background contrast, and high resolution, have been obtained. The detailedcontents are as follow:1. A new path that combines the terahertz imaging principles, image de-noising andenhancement techniques has been put forward to implement terahertz imagede-noising and enhancement process. In order to eliminate the effect ofexperimental noise, laser power jitter and terahertz radiation fluctuation onterahertz images, methods of image de-noising and enhancement have beenintegrated in spatial domain and frequency domain. The definition andidentifiability of terahertz images have been improved effectively. The effects ofdifferent de-noising and enhancement methods have also been analyzed.2. The physical model of homomorphic filtering for terahertz imaging to suppressthe fluctuation effect has been established. Through the modified homomorphicfiltering for the original pulsed terahertz images and continuous-wave terahertzimages, and the comparison with the processing effects of previous literaturequantitatively, the conclusion has been demonstrated that, our method can suppress the background fluctuation effect and enhance the imaging targetssimultaneously, both in pulsed and CW terahertz imaging.3. The high resolution reconstruction of terahertz images has been implementedtaking advantage of super-resolution image processing techniques. Theobservation model of high resolution image reconstruction and the prerequisite ofhigh resolution reconstruction-subpixel shift, have been illustrated. In order tocomplete high resolution reconstruction of terahertz images, the methodsincluding Lucy-Richardson iteration, projection onto convex sets approach(POCS), iterative back projection (IBP) SR reconstruction approach, waveletdecomposition method and interpolation method have been used.The research in this dissertation provides valuable theoretical analysis andexperimental basis for high resolution terahertz imaging and image processing.Furthermore, it also indicates the direction of high resolution terahertz imaging in thenear future.