本文選題：紅外目標(biāo)檢測(cè)與識(shí)別　切入點(diǎn)：主動(dòng)輪廓模型　出處：《天津理工大學(xué)》2017年碩士論文

【摘要】：近年來(lái),隨著紅外熱成像系統(tǒng)的快速發(fā)展,紅外圖像目標(biāo)檢測(cè)與識(shí)別技術(shù)已經(jīng)成為現(xiàn)代圖像處理領(lǐng)域的重要研究課題,在軍事、醫(yī)學(xué)、監(jiān)控、交通等領(lǐng)域都發(fā)揮著重大的作用。本文分析了紅外圖像目標(biāo)檢測(cè)與識(shí)別技術(shù)中需要解決的關(guān)鍵問題,在紅外圖像目標(biāo)檢測(cè)和目標(biāo)識(shí)別方面進(jìn)行了深入研究。首先對(duì)常用的紅外圖像目標(biāo)檢測(cè)算法進(jìn)行研究,主要研究了邊緣檢測(cè)算法、區(qū)域生長(zhǎng)法和主動(dòng)輪廓模型,并對(duì)算法進(jìn)行了仿真實(shí)現(xiàn);然后針對(duì)主動(dòng)輪廓模型存在的對(duì)初始輪廓位置敏感、凹性區(qū)域輪廓無(wú)法正確收斂等問題,設(shè)計(jì)了一種將自適應(yīng)邊緣檢測(cè)與主動(dòng)輪廓模型相融合的紅外目標(biāo)自主檢測(cè)算法。仿真實(shí)驗(yàn)與分析證實(shí),該算法可以實(shí)現(xiàn)紅外目標(biāo)輪廓的全自動(dòng)提取,能夠增強(qiáng)初始輪廓精確收斂到真實(shí)邊界的能力,提高檢測(cè)精度。然后在紅外圖像目標(biāo)識(shí)別方面,針對(duì)幾何不變矩和Hu氏不變矩等特征提取方法進(jìn)行研究,并對(duì)其穩(wěn)定性和可區(qū)分性進(jìn)行深入分析;針對(duì)最小距離分類器和Bayes分類器進(jìn)行研究,將幾何不變矩和Hu氏不變矩與最小距離分類器和Bayes分類器進(jìn)行仿真分析,并提出了一種基于Bayes網(wǎng)絡(luò)的紅外目標(biāo)識(shí)別算法。實(shí)驗(yàn)結(jié)果表明,基于Bayes網(wǎng)絡(luò)的紅外目標(biāo)識(shí)別算法的識(shí)別率高,有較好的分類識(shí)別效果。最后,基于VS2010和OpenCV2.4.3搭建的軟件平臺(tái),實(shí)現(xiàn)了紅外圖像目標(biāo)自主檢測(cè)算法,并通過實(shí)驗(yàn)驗(yàn)證算法的可行性,實(shí)驗(yàn)結(jié)果表明本文提出的紅外圖像目標(biāo)自主檢測(cè)算法可以實(shí)現(xiàn)紅外目標(biāo)輪廓的精確自動(dòng)收斂。
[Abstract]:In recent years, with the rapid development of infrared thermal imaging system, infrared image target detection and recognition technology has become an important research topic in the field of modern image processing, in military, medicine, monitoring, Traffic and other fields play an important role. In this paper, the key problems in infrared image target detection and recognition technology are analyzed. Firstly, the common algorithms of infrared image target detection are studied, including edge detection algorithm, region growth method and active contour model. Then the active contour model is sensitive to the initial contour position and the concave region contour can not converge correctly. An autonomous infrared target detection algorithm combining adaptive edge detection and active contour model is designed. The simulation results and analysis show that the algorithm can automatically extract the contour of infrared target. It can enhance the ability of the initial contour to converge to the real boundary and improve the detection accuracy. Then, in infrared image target recognition, the methods of feature extraction, such as geometric invariant moment and Hu's invariant moment, are studied. The stability and distinguishability of the classifier are analyzed deeply, and the geometric invariant moment, Hu's invariant moment and the minimum distance classifier and Bayes classifier are simulated and analyzed, aiming at the minimum distance classifier and Bayes classifier. An infrared target recognition algorithm based on Bayes network is proposed. The experimental results show that the infrared target recognition algorithm based on Bayes network has high recognition rate and good classification and recognition effect. Finally, the software platform based on VS2010 and OpenCV2.4.3 is built. The feasibility of the algorithm is verified by experiments. The experimental results show that the proposed algorithm can achieve accurate automatic convergence of infrared image contour.
【學(xué)位授予單位】：天津理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 湯澤瀅;基于模糊規(guī)則自動(dòng)獲取的模糊主動(dòng)輪廓模型[J];蘇州大學(xué)學(xué)報(bào)(自然科學(xué)版);2005年03期

2 方偉;陳會(huì)勇;陳宗海;;基于非參數(shù)二維熵的主動(dòng)輪廓模型[J];模式識(shí)別與人工智能;2005年06期

3 趙明珠;陳勝勇;管秋;;基于混合主動(dòng)輪廓模型和區(qū)域間差別最大化的細(xì)胞弱邊界分割[J];計(jì)算機(jī)應(yīng)用與軟件;2011年11期

4 張寧;余學(xué)飛;;基于加性算子分割的快速靜磁場(chǎng)主動(dòng)輪廓模型[J];計(jì)算機(jī)工程與應(yīng)用;2012年05期

5 陳會(huì)勇;胡玉鎖;陳宗海;;基于恒定曲率變化的主動(dòng)輪廓模型[J];中國(guó)圖象圖形學(xué)報(bào);2006年06期

6 李小毛;王智峰;唐延?xùn)|;;基于形狀保持主動(dòng)輪廓模型長(zhǎng)直條的檢測(cè)[J];計(jì)算機(jī)工程;2008年01期

7 陶玲;錢志余;陳春曉;;主動(dòng)輪廓模型及其在醫(yī)學(xué)體分割中的應(yīng)用[J];華南理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2008年01期

8 李燕;羅四維;鄒琪;;帶評(píng)價(jià)系統(tǒng)的曲率相關(guān)有向主動(dòng)輪廓模型[J];北京交通大學(xué)學(xué)報(bào);2010年02期

9 岑峰,戚飛虎,曾文s，

本文編號(hào)：1656183