本文選題：紅外焦平面陣列 + 非均勻性校正　； 參考：《南京理工大學》2016年碩士論文

【摘要】：近年來,紅外焦平面陣列作為紅外熱成像的核心技術(shù)得到了長足地發(fā)展,其以靈敏度高、可靠性強的特性在航空、工業(yè)、農(nóng)業(yè)、醫(yī)學和科學研究等領(lǐng)域廣泛應(yīng)用。然而,由于制造工藝的限制,使得其焦平面的光敏元輸出響應(yīng)不同,導致圖像存在著大量非均勻性噪聲,這大大制約了紅外熱成像系統(tǒng)的應(yīng)用與研究。所以對非均勻性校正技術(shù)的研究刻不容緩。本文對非均勻校正技術(shù)深入研究,實現(xiàn)了一種基于FPGA平臺的非均勻校正系統(tǒng)。本文首先就非均勻性的產(chǎn)生原因作了簡單的闡述,并介紹了幾種常用的非均勻性校正方法。然后結(jié)合配準算法并遵行利于FPGA實現(xiàn)的原則,提出了一種基于模板匹配的非均勻性校正算法。該算法選取圖像中的某一區(qū)域作為基準模塊,利用菱形搜索模式,得到幀間圖像的偏移信息,并利用最陡下降法求得校正的增益和偏置系數(shù)。由于配準類算法產(chǎn)生鬼影的原因大都是匹配不準的問題,因此在算法中引入了鬼影消除的機制。經(jīng)過實驗仿真及FPGA硬件實時校正,基于模板匹配的非均勻性校正算法實現(xiàn)了幀間圖像的實時快速匹配,并在收斂速度和校正效果上都有顯著地成效,且有效地抑制了鬼影的產(chǎn)生。
[Abstract]:In recent years, infrared focal plane array (IRFPA), as the core technology of infrared thermal imaging, has been widely used in aviation, industry, agriculture, medicine and scientific research due to its high sensitivity and reliability. However, due to the limitation of manufacturing process, the output response of Guang Min element in the focal plane is different, which leads to a large number of non-uniform noise in the image, which greatly restricts the application and research of infrared thermal imaging system. Therefore, the study of non-uniformity correction technology is urgent. In this paper, a non-uniformity correction system based on FPGA platform is realized by deeply studying the non-uniformity correction technology. In this paper, the causes of nonuniformity are briefly described, and several commonly used nonuniformity correction methods are introduced. Then a template matching based nonuniformity correction algorithm is proposed based on the registration algorithm and the principle of FPGA implementation. The algorithm selects a certain region of the image as the reference module, uses the rhombus search mode to obtain the offset information of the inter-frame image, and uses the steepest descent method to obtain the corrected gain and offset coefficient. Due to the problem of mismatch, the mechanism of ghost elimination is introduced in the registration algorithm. After experimental simulation and FPGA hardware real-time correction, the non-uniformity correction algorithm based on template matching is used to realize real-time and fast image matching between frames, and the convergence speed and correction effect are remarkable. And effectively inhibited the generation of ghosts.
【學位授予單位】：南京理工大學
【學位級別】：碩士
【學位授予年份】：2016
【分類號】：TN219;TP391.41
，

本文編號：1851869