چكيده به لاتين
One of the applications of Passive Millimeter-Wave Imaging (PMMW) is to show hidden objects. Indicating hidden objects in the places such as airports due to security is extremely important. Millimeter wave images, due to its received wave’s characteristics, have the ability to detect hidden objects which are under people clothes, but its receiver antenna limitation that operates like a low-pass filter causes to provide low-quality images from the scene. Using image processing techniques is one of the methods to improve the quality of these images. Images combining and recovering are two proposed techniques in image processing to enhance these images quality. In image recovering, it is trying to use Point Spread Function (PSF) estimation, which is image blurring factor, to recover original image reverse. Most of the proposed algorithms for this purpose use repetitive algorithms. Image combining is one of the techniques to improve image quality when such image is obtained with various characteristic from one scene. In these methods, images from different sources are combined to make a compound image contains all input images information.
in this thesis, first different methods of image retrieval and combining for millimeter wave images are investigated and implemented and then some of the appropriate approaches for more works in both fields as for the application are proposed. Applying the methods and processes methods, there is a significant upgrade in images visual quality, while to detect hidden objects and images retrieval and combination a process is proposed that cause significant improvement in the desired application. Also, a proper process to enhance passive millimeter wave images quality to detect hidden objects is discussed and design considerations of such process are proposed. According to implementation results and study and super-resolution and fusion methods, it is expected that there is an improvement around 20% recoveries in the overall process according to the intended application for SSIM benchmark, which is related to image structure maintain.
In this thesis, a new method is used to enhance passive millimeter wave images resolution based on pseudo-blind deconvolution that actually PSF filter information is available and the purpose is to improve image noise and blur. To this, a new relationship has been defined that by minimizing it using descent gradient algorithm, the output image will have a better quality than other methods. Output image improvement rate with PSNR and SSIM standard features has been researched. In Lena and cameraman image respectively up to 4.62 and 3.74 PSNR units for low noise is made and also the significant upgrade in output images visual quality is observed. The proposed method gives up to 6.2% improvement than last methods in this field. Also, quantity evaluation of images passive millimeter wave because of reference images absence was not possible before, so that a benchmark without appropriate reference is suggested and by observing upgraded images with the proposed algorithm, the efficiency algorithm is clearly visible. This improvement is visible in terms of becoming more specific in details of millimeter wave image instead losing little quality in image corners that don’t have diagnostic importance. In this thesis, a radiometer system including low noise amplifiers, local mixer and oscillator, Ku band detector, various band filters, and data receiving and processing system has been designed that reaching maximum gain and minimum noise is possible. The aim of this design was to achieve minimum noise equivalent temperature difference for the whole system by adjusting and analyzing all design free parameters. As a result of this design, each Kelvin variation in the input, gives 2.51 μv variations in output voltage and this design provides the optimal construction of passive millimeter wave operation system. In continue, to evaluate the designed system, output the image, which has been obtained from the designed system, is formed. After forming the image, due to inherent blur property of millimeter wave images, modification and improvement process of image visual quality is performed on the obtained image. In the images processing section, an effective algorithm for super-resolution, fusion, and their applying arrangement is proposed. The mentioned methods are applied to output and simulated images and caused improvement in output images.
