شماره ركورد
25593
شماره راهنما
ELE3 86
عنوان
ﺑﻬﺒﻮﺩﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ ﺑﺎ ﺭﻭﻳﻜﺮﺩ ﭼﻨﺪﻣﻘﻴﺎﺳﻲ ﻭ ﻛﺎﻫﺶ ﺍﺛﺮ ﭘﺲ ﺯﻣﻴﻨﻪﻫﺎﻱ ﺳﺎﺧﺘﺎﺭﻱ
مقطع تحصيلي
دكتري
رشته تحصيلي
مهندسي برق - الكترونيك
دانشكده
فني و مهندسي
تاريخ دفاع
1402/09/04
صفحه شمار
105 ص.
استاد راهنما
پيمان معلم , محمد كاظمي
استاد مشاور
ﺩﻛﺘﺮ ﺳﺎﻋﺪ ﻣﺮﺍﺩﻱ
كليدواژه فارسي
ﺳﺎﻣﺎﻧﻪ ﻫﺎﻱ ﺟﺴﺘﺠﻮ ﻭ ﺭﺩﻳﺎﺑﻲ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ , ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﻫﺪﻑ , ﺩﺭﻭﻥ ﻳﺎﺑﻲ ﺟﻬﺘﻲ , ﺍﻟﮕﻮﻱ ﺭﻳﺨﺖ ﺷﻨﺎﺳﻲ , ﻣﻌﻴﺎﺭُﺗُﻨﻚ ﺑﻮﺩﻥ
چكيده فارسي
ﺳﺎﻣﺎﻧﻪ ﻫﺎﻱ ﺟﺴﺘﺠﻮ ﻭ ﺭﺩﻳﺎﺑﻲ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ ﺍﺑﺰﺍﺭﻱ ﺟﻬﺖ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﺗﻬﺎﺟﻤﻲ ﺩﺭ ﺗﺼﺎﻭﻳﺮ ﻣﺎﺩﻭﻥﻗﺮﻣﺰ ﻫﺴﺘﻨﺪ. ﻛﺎﺭﺑﺮﺩ ﺍﻳﻦ ﺍﺩﻭﺍﺕ ﺩﺭ ﺻﻨﺎﻳﻊ ﺩﻓﺎﻋﻲ ﺑﻪ ﺩﻟﻴﻞ ﺧﺼﻮﺻﻴّﺎﺗﻲ ﻫﻤﭽﻮﻥ ﻗﺎﺑﻠﻴّﺖ ﺍﺧﺘﻔﺎ، ﻣﻘﺎﻭﻣﺖ ﺩﺭ ﺑﺮﺍﺑﺮ ﺍﻏﺘﺸﺎﺵ ﻭ ﺍﺧﺘﻼﻝ ﻭ ﺗﻮﺍﻧﺎﻳﻲ ﺩﺭ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﺑﺎ ﺳﻄﺢ ﻣﻘﻄﻊ ﺭﺍﺩﺍﺭﻱ ﻛﻢ، ﺍﻓﺰﺍﻳﺶ ﭼﺸﻤﮕﻴﺮﻱ ﻳﺎﻓﺘﻪ ﺍﺳﺖ.
ﺑﺎ ﺗﻮﺟّﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﺍﻫﺪﺍﻑ ﺑﺎﻳﺪ ﺩﺭ ﺩﻭﺭﺗﺮﻳﻦ ﻓﺎﺻﻠﻪ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺷﻮﻧﺪ، ﻭ ﺍﺯ ﻃﺮﻓﻲ ﺩﻳﮕﺮ ﺗﺼﺎﻭﻳﺮ ﻣﺎﺩﻭﻥﻗﺮﻣﺰ ﻣﻌﻤﻮﻻً ﺁﻏﺸﺘﻪ ﺑﻪ ﻧﻮﻳﺰ ﺑﻮﺩﻩ ﻭ ﺩﺍﺭﺍﻱ ﭘﺲﺯﻣﻴﻨﻪ ﻱ ﭘﻴﭽﻴﺪﻩ ﻫﺴﺘﻨﺪ، ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﺩﺭ ﺍﻳﻦ ﺳﺎﻣﺎﻧﻪﻫﺎ ﺍﻣﺮﻱ ﭼﺎﻟﺶ ﺑﺮﺍﻧﮕﻴﺰ ﺍﺳﺖ. ﻣﺸﺨّﺺ ﻧﺒﻮﺩﻥ ﺍﻧﺪﺍﺯﻩ ﻱ ﻫﺪﻑ ﺍﺯ ﺩﻳﮕﺮ ﭼﺎﻟﺶ ﻫﺎﻱ ﻣﻮﺟﻮﺩ ﺑﺮ ﺳﺮ ﺭﺍﻩ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺁﻥ ﻣﻲ ﺑﺎﺷﺪ. ﻫﻤﻴﻨﻄﻮﺭ، ﻗﺮﺍﺭﮔﻴﺮﻱ ﻫﺪﻑ ﺩﺭ ﻣﺠﺎﻭﺭﺕ ﻫﺪﻑ ﻧﻤﺎﻱ ﺑﺰﺭﮒ ﻣﻲ ﺗﻮﺍﻧﺪ ﻛﻴﻔﻴﺖ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺭﺍ ﺑﻪ ﺷﺪّﺕ ﻣﺨﺪﻭﺵ ﻛﻨﺪ.
ﺍﺯ ﻫﻤﻴﻦ ﺭﻭﻱ، ﺩﺭ ّﺍﻭﻟﻴﻦ ﮔﺎﻡ ﺍﺯ ﺍﺟﺮﺍﻱ ﺍﻳﻦ ﭘﮋﻭﻫﺶ، ﺍﻧﺪﺍﺯﻩ ﻱ ﭘﻨﺠﺮﻩﻱ ﭘﺮﺩﺍﺯﺵ ﻣﻄﺎﻟﻌﻪ ﺷﺪﻩ ﻭ ﺑﻪ ﻭﺳﻴﻠﻪﻱ ﻣﺒﺎﻧﻲ ﺗﺌﻮﺭﻱ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ، ﺍﻧﺪﺍﺯﻩ ﻭ ﺷﻜﻞ ﻓﻴﻠﺘﺮ ﺑﻬﻴﻨﻪ ﺑﺮﺍﻱ ﭘﺮﺩﺍﺯﺵ ﺗﺼﺎﻭﻳﺮ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ ﻣﻮﺭﺩ ﺑﺮﺭﺳﻲ ﻗﺮﺍﺭ ﻣﻲ ﮔﻴﺮﺩ. ﺍﻳﻦ ﻣﺴﺌﻠﻪ ﺑﻪ ﺩﻟﻴﻞ ﻧﺎﻣﺸﺨﺺ ﺑﻮﺩﻥ ﺍﻧﺪﺍﺯﻩ ﻱ ﻫﺪﻑ ﺍﻫﻤّﻴﺖ ﺯﻳﺎﺩﻱ ﺩﺍﺭﺩ.
ﺩﺭ ﮔﺎﻡ ﺑﻌﺪﻱ، ﺑﺎ ﺗﻤﺮﻛﺰ ﺑﺮ ﺍﻫﺪﺍﻑ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﺪﻑ ﻧﻤﺎ، ﺳﻪ ﺭﻭﺵ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ ﺍﺭﺍﺋﻪ ﻣﻲﺷﻮﻧﺪ
ﻭ ﺩﺭ ﺻﺤﻨﻪ ﻫﺎﻱ ﻭﺍﻗﻌﻲ ﻣﺨﺘﻠﻒ ﺍﺯ ﻧﻈﺮ ﻛﻤﻲ ﻭ ﻛﻴﻔﻲ ﺑﺮﺭﺳﻲ ﻣﻲ ﺷﻮﻧﺪ. ﺩﺭ ﺭﻭﻳﻜﺮﺩ ّﺍﻭﻝ، ﺍﺯ ﺩﺭﻭﻥ ﻳﺎﺑﻲ ﺟﻬﺘﻲ ﺍﺳﺘﻔﺎﺩﻩ ﺷﺪﻩ ﺍﺳﺖ. ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ ﺍﻳﻦ ﻣﺴﺌﻠﻪ ﺗﻮﺟﻪ ﺩﺍﺭﺩ ﻛﻪ ﺍﺟﺰﺍﻱ ﺗﺼﻮﻳﺮ ﻛﻪ ﺟﺰﺋﻴﺎﺕ ﺑﻴﺸﺘﺮﻱ ﺩﺍﺭﻧﺪ، ﺳﺎﺧﺘﺎﺭ ﻫﻨﺪﺳﻲ ﻳﺎ ﺍﻟﮕﻮﻱ ﻣﺸﺨﺼﻲ ﺩﺍﺭﻧﺪ. ﺍﻳﻦ ﻣﻮﺿﻮﻉ ﺩﺭ ﻟﺒﻪ ﻫﺎﻱ ﺗﺼﻮﻳﺮ ﻣﺸﻬﻮﺩ ﻣﻲ ﺑﺎﺷﺪ. ﺩﺭ ﭼﻨﻴﻦ ﻣﻮﺍﺭﺩﻱ ﺩﺭﻭﻥﻳﺎﺑﻲ ﺩﺭ ﺟﻬﺖ ﻓﺮﻛﺎﻧﺴﻲ ﭘﺎﻳﻴﻦ، ﻳﺎ ﺑﻪ ﻋﺒﺎﺭﺗﻲ ﻫﻤﺴﻮ ﺑﺎ ﻟﺒﻪ ﻫﺎ ﻧﺘﺎﻳﺞ ﺑﻬﺘﺮﻱ ﻧﺴﺒﺖ ﺑﻪ ﺩﺭﻭﻥﻳﺎﺑﻲ ﺩﺭ ﺧﻼﻑ ﺟﻬﺖ ﻟﺒﻪ ﻣﻲ ﺩﻫﺪ. ﺩﺭ ﺭﻭﺵ ﺩﻭﻡ، ﻳﻚ ﺍﻟﮕﻮﻱ ﺭﻳﺨﺖ ﺷﻨﺎﺳﻲ ﭼﻨﺪ ﻃﺒﻘﻪ ﺍﺭﺍﺋﻪ ﻣﻲ ﺷﻮﺩ. ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ، ﺗﻮﺟّﻪ ﻭﻳﮋﻩ ﺍﻱ ﺑﻪ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﺍﻫﺪﺍﻑ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﺪﻑ ﻧﻤﺎ ﺷﺪﻩ ﺍﺳﺖ. ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺍﻋﻤﺎﻝ ﺍﻳﻦ ﺭﻭﻳﻜﺮﺩ ﺑﻪ ﺗﺼﺎﻭﻳﺮ ﻣﺎﺩﻭﻥ ﻗﺮﻣﺰ ﻭﺍﻗﻌﻲ ﻗﺎﺑﻠﻴّﺖ ﺑﺎﻻﻱ ﺍﻳﻦ ﺭﻭﻳﻜﺮﺩ ﺩﺭ ﺍﻓﺰﺍﻳﺶ ﻛﺎﺭﺍﻳﻲ ﺳﺎﻣﺎﻧﻪ ﻱ ﺟﺴﺘﺠﻮﮔﺮ ﺭﺍ ﻧﺸﺎﻥ ﻣﻲ ﺩﻫﺪ. ﻭ ﺩﺭ ﻧﻬﺎﻳﺖ، ﺩﺭ ﺭﻭﻳﻜﺮﺩ ﺳﻮﻡ، ﺍﺯ ﺑﺮﺩﺍﺭ ﺷﻴﺐ ﻣﺤﻠﻲ ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﺑﺮﺩﺍﺭ ﻭﻳﮋﮔﻲ ﺑﺮﺍﻱ ﺁﺷﻜﺎﺭﺳﺎﺯﻱ ﻫﺪﻑ ﺍﺳﺘﻔﺎﺩﻩ ﺷﺪﻩ ﺍﺳﺖ. ﻧﺸﺎﻥ ﻣﻲ ﺩﻫﻴﻢ ﻛﻪ ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﻭﻳﮋﮔﻲ ﺑﺮﺍﻱ ﭘﺲ ﺯﻣﻴﻨﻪ ﺗﻨﻚ ﺗﺮ ﺍﺯ ﻫﺪﻑ ﻣﻲﺑﺎﺷﺪ، ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﻳﻚ ﻣﻌﻴﺎﺭ ﺗﻨﻚ ﻣﻲ ﺗﻮﺍﻥ ﻫﺪﻑ ﺭﺍ ﺍﺯ ﭘﺲ ﺯﻣﻴﻨﻪ ﺗﻤﻴﺰ ﺩﺍﺩ. ﻧﺘﺎﻳﺞ ﺷﺒﻴﻪﺳﺎﺯﻱﻫﺎ ﺣﺎﻛﻲ ﺍﺯ ﺑﺮﺗﺮﻱ ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎﻱ ﭘﻴﺸﻨﻬﺎﺩﻱ ﺩﺭ ﺗﻘﻮﻳﺖ ﻧﺎﺣﻴﻪ ﻱ ﻫﺪﻑ ﻭ ﺗﻀﻌﻴﻒ ﻧﺎﺣﻴﻪ ﻱ ﭘﺲ ﺯﻣﻴﻨﻪ ﺩﺭ ﺷﺮﺍﻳﻂ ﻭﺍﻗﻌﻲ ﻣﺨﺘﻠﻒ ﻣﻲ ﺑﺎﺷﺪ. ﺑﻪ ﻃﻮﺭ ﻣﻴﺎﻧﮕﻴﻦ ﻧﺴﺒﺖ ﺳﻴﮕﻨﺎﻝ ﺑﻪ ﻧﻮﻳﺰ ﺗﺼﺎﻭﻳﺮ ﺧﺮﻭﺟﻲ ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎﻱ ﭘﻴﺸﻨﻬﺎﺩﻱ ﺗﺎ ٪06 ﻋﻤﻠﻜﺮﺩ ﺑﻬﺘﺮﻱ ﻧﺴﺒﺖ ﺑﻪ ﺍﻟﮕﻮﺭﻳﺘﻢ ﻫﺎﻱ ﻣﻮﺭﺩ ﻣﻘﺎﻳﺴﻪ ﺩﺍﺭﺩ.
كليدواژه لاتين
Infrared search and track system , Target detection , Directional Interpo-lation , Morphological operations , Sparsity measure
عنوان لاتين
Structural background suppression to improve infrared target detection ability using multi-scale approach
گروه آموزشي
مهندسي برق
چكيده لاتين
An infrared search and track (IRST) system is an instrument for detecting and track-ing incoming threats which give off infrared radiations. The application of these devices grow vastly due to special characteristics like covert operation, robustness against jamming and good capability in detecting low radar cross section objects.
Since the incoming threats should be detected as away as possible from the searcher system, the resulting corresponding image of the small infrared target on the image plane is a dim, small, and hard-to-detect spot. Moreover, noisy and complicated background in infrared scenarios make situation much worse. On the other hand, when a target is close to a clutter, the detection performance highly degrades.
In the first step of implementing research plan, the size and shape of processing ker-nel is deeply studied based on the detection theory principles. The importance of kernel size size is heightened due to unknown target size.
Thereafter, considering the challenges of detecting a target positioned in the vicin-ity of a background, three detection algorithms are developed. The first method employs a directional interpolation schema. This method first analyses the local structure of the image and then, performs the background estimation based on the minimum directional interpolation. In the second work, a multi-stage detector based on morphological operations is proposed. ُThis approach particularly focuses on de-tecting targets that are in close proximity to clutter. Finally, a local gradient vector is developed as a feature vector for the purpose of infrared target detection. We show that this feature vector is sparser in the background region than the target region, therefore, a sparsity measure can be utilized to distinguish between a target and it’s background. Simulation results demonstrate improvement in target detection abil-ity in the proposed algorithms. On average, the proposed algorithms improves the output signal-to-noise ration by 60% compared to the other algorithms.
تعداد فصل ها
5
فهرست مطالب pdf
154370
نويسنده