Research On Enhancement And Segmentation Of Passive Millimeter Wave (PMMW) Image

Passive millimeter wave(PMMW)imaging system achieves imaging through the difference of millimeter energy that objects radiate,which has great performance in penetrating and excellent ability to recognize metal target from surrounding environment.These features enable it to have a wide application in the fields like battlefield reconnaissance and security checking of important places.Even if the scanning technology of PMMW imaging system has undergone tremendous improvements in acquisition speed and signal-to-noise ratio,PMMW images still hampered with degradations like dislocation,noise and other artifacts.Those factors have adverse effects on subsequent image processing of millimeter image such as feature extraction,registration,fusion,and super resolution etc.Therefore,it is essential to propose a practical algorithm for PMMW image pre-processing.This paper aims to study passive millimeter wave pre-processing technology,the main contents include the following aspects:(1)The theory of passive millimeter wave(PMMW)imaging system is studied,the effects of different scanning mechanisms on the results is analyzed.(2)Traditional PDEs based algorithm cannot remove the noise effectively if the image degradation deeply.Thus,a novel scheme for PMMW image de-nosing based on a PDEs model was proposed in this paper to avoid blocky effects and speckle artifacts while achieving good trade-off between noise removal and edge preservation.The experimental results show the algorithms performance well.(3)The PMMW image processed by traditional PDEs based methods tends to generate multiple false edges or enhance noise because of the limit of the single channel device.So,we proposed noise removal algorithm based on local PDES processing.The experimental results show that algorithm work well.(4)Study the noise model for multi-channel passive millimeter wave(PMMW)imaging system.Then,a de-nosing algorithm based on image morphology and PDEs was proposed.The experimental results show the algorithms work effectively.(5)The traditional image segmentation algorithm is analyzed.Because traditional threshold based method performance bad on the PMMW image,we use the edge information to improve the algorithms based on threshold.The experimental results show the algorithms effectively meet the needs of the project.