本文選題：視頻圖像序列 + 邊緣檢測��； 參考：《沈陽理工大學》2017年碩士論文

【摘要】：視頻圖像序列中的運動目標跟蹤一直是計算機領域中的熱點問題,在動態(tài)場景中運動目標的檢測和跟蹤技術通常可以利用在視頻監(jiān)控、人機交互、汽車輔助駕駛、運動行為分析等方面。在實際的生活生產過程之中,目標跟蹤技術仍然是計算機視覺中富有挑戰(zhàn)性的工作。基于視頻圖像序列中運動目標的特征和其跟蹤方法和應用,本文作出工作如下:(1)文中提出一種新的邊緣檢測方法用來改善圖像邊緣檢測噪聲點過多的問題。圖像的邊緣被用來進行圖像分割、模版匹配和圖像識別等方面的探討和研究,是圖像最基本的特征和構成因素。在圖像的邊緣檢測內容中,由于檢測圖像邊緣的方法式多樣化的,但其中適用性最廣最為快速的方法還是基于模糊梯度算法中的邊緣檢測方式。文中對現有的模糊梯度算法進行評估和討論,希望能夠解決掉其中存在的缺點問題并且對之進行進一步的完善,最終得到一種更新型更好的圖像邊緣檢測方法。(2)在基于現有的運動目標跟蹤方法中,對于圖像的復雜背景有噪聲干擾的情況下,改善幀間相減法算法來是目標跟蹤方法更加完善。很多物體都會隨著時間而進行運動,在實際生活中,不同的運動物體對于不同的群體傳達著非常不一樣的視覺信息,而這些視覺信息又會給群體們帶來非常重要的現實意義,人們用視覺所捕捉到的信息往往會對其有實際意義與使用價值。在研究基于視頻圖像的運動目標的檢測與追蹤具有著重大意義,目前實驗中已經實現了對運動物體的檢測與追蹤,在探究理論的過程中,發(fā)現幀間差分法在獲取運動目標并跟蹤的這種形式是最為簡便和完整可靠的一種。在運動目標追蹤檢測方面,運用的復雜度的分塊搜索法來進行計算,能夠將其進行仿真模擬實驗與編程實現。(3)針對圖像在一些光暗程度變化較大時和圖像的背景復雜的情況下本文改善兩種圖像清晰化算法,以及和原有的卡爾曼濾波方法進行比對。在公共環(huán)境的安全和防護成為人們關注的熱點問題同時,而視頻監(jiān)控即為最常用的公共安全保護措施,在馬路、銀行、醫(yī)院、學校等公共區(qū)域,視頻監(jiān)控技術被廣泛應用與生產生活之中。雖然高清攝像頭已經逐漸開始普及,在圖像受到一定的光暗程度變化干擾和復雜背景時會使原有的算法沒有明顯效果,本文使用維納濾波算法和同態(tài)濾波算法來進行圖像的清晰化工作。
[Abstract]:Moving target tracking in video sequence has always been a hot issue in the computer field. Detection and tracking techniques of moving targets in dynamic scenes are usually used in video surveillance, human-computer interaction, vehicle driving, and motion behavior analysis. In the actual production process, target tracking technology is still the same. A challenging job in computer vision. Based on the features of the moving target in the video sequence and its tracking methods and applications, the following work is made as follows: (1) a new edge detection method is proposed to improve the problem of excessive noise points in the image edge detection. The edge of the image image is used for image segmentation and template matching. Research and Research on image recognition are the most basic features and components of images. In the content of image edge detection, the method of detecting the edge of the image is diversified, but the most widely and most fast method is based on the edge detection method in the fuzzy gradient algorithm. The algorithm is evaluated and discussed. We hope to solve the shortcomings of the existing problems and further improve it, and finally get a new and better image edge detection method. (2) in the current moving target tracking method, the interframe subtraction is improved for the complex background of the image with noise interference. The algorithm is that the target tracking method is more perfect. Many objects will move along with time. In real life, different moving objects convey very different visual information to different groups, and these visual information will bring very important realistic meaning to the group, people use the information captured by the vision. It is of great significance for the detection and tracking of moving objects based on video images. At present, the detection and tracking of moving objects have been realized in the experiment. In the process of exploring the theory, it is found that the form of inter frame difference method in obtaining moving targets and tracking is the most simple form. It is a complete and reliable one. In the field of tracking and detection of moving targets, the complexity of the block search method is used to carry out the calculation. (3) to improve the image sharpening algorithm, two kinds of image sharpening algorithms are improved in this paper, when the light and dark degree of the image is varied and the background of the image is complex. Compared with the original Calman filtering method, the security and protection of the public environment have become the hot issues of attention, and video surveillance is the most common public security protection measures, in the public areas such as roads, banks, hospitals, schools and other public areas. Video surveillance technology is widely used and in production life. It has gradually become popular. When the image is disturbed by a certain degree of light and dark, the original algorithm has no obvious effect. In this paper, the Wiener filtering algorithm and homomorphic filtering algorithm are used to make the image clear.

【學位授予單位】：沈陽理工大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41

【參考文獻】

相關期刊論文 前10條

1 劉洋;;交通樞紐智能交通系統(tǒng)設計方案分析[J];科技創(chuàng)新與應用;2013年13期

2 李秋林;何家峰;;基于三幀差法和交叉熵閾值法的車輛檢測[J];計算機工程;2011年04期

3 趙鵬軍;劉三陽;;一種新的智能優(yōu)化及其改進研究[J];小型微型計算機系統(tǒng);2010年05期

4 張振東;;智能交通系統(tǒng)概述及國內外發(fā)展狀況[J];科學之友;2010年06期

5 連曉峰;張_";戴軍;劉載文;王小藝;;基于視頻流的運動人體檢測方法研究[J];北京工商大學學報(自然科學版);2009年06期

6 龔振宇;龐全;范影樂;;自適應隨機共振的圖像復原研究[J];計算機工程與科學;2009年05期

7 沈峰亭;魏紅;;基于改進Sobel算子的螺紋邊緣檢測[J];微計算機信息;2008年01期

8 夏克寒;許化龍;張樸睿;;粒子濾波的關鍵技術及應用[J];電光與控制;2005年06期

9 馬艷,張治輝;幾種邊緣檢測算子的比較[J];工礦自動化;2004年01期

10 王亮,胡衛(wèi)明,譚鐵牛;人運動的視覺分析綜述[J];計算機學報;2002年03期

相關博士學位論文 前1條

1 姚劍敏;粒子濾波跟蹤方法研究[D];中國科學院研究生院（長春光學精密機械與物理研究所）;2005年

相關碩士學位論文 前2條

1 王奇;基于視頻圖像的交通監(jiān)控系統(tǒng)研究[D];西北工業(yè)大學;2005年

2 許春香;圖像增強技術在煤礦視頻監(jiān)控系統(tǒng)中的應用與研究[D];鄭州大學;2003年

，

本文編號：1885754