發(fā)布時(shí)間：2018-06-09 22:06

  本文選題：運(yùn)動(dòng)目標(biāo)檢測(cè) + 陰影消除　； 參考：《北京工業(yè)大學(xué)》2015年碩士論文

【摘要】：高速公路是城市快速交通的主干線,已成為經(jīng)濟(jì)發(fā)展的脊梁。高速公路路面狀態(tài)的安全級(jí)別受自然因素影響較大,比如降雨、大霧、降雪與融雪等都將影響高速公路的運(yùn)營(yíng)狀況。本文以高速公路智能化視頻監(jiān)控系統(tǒng)開(kāi)發(fā)為背景,研究高速公路監(jiān)控視頻中運(yùn)動(dòng)目標(biāo)識(shí)別、運(yùn)動(dòng)車輛跟蹤和道路雨雪狀態(tài)檢測(cè)技術(shù),并在基礎(chǔ)上完成了應(yīng)用系統(tǒng)設(shè)計(jì)和功能模塊開(kāi)發(fā)。本文主要工作包括:1)提出了一種新的目標(biāo)陰影消除方法。為了準(zhǔn)確檢測(cè)運(yùn)動(dòng)車輛目標(biāo),需要首先進(jìn)行運(yùn)動(dòng)陰影消除。針對(duì)傳統(tǒng)HSV顏色空間下陰影消除方法的不足,本文提出了一種HSV顏色空間下結(jié)合圖像區(qū)域信息的陰影消除方法。使用該方法可以避免因車輛目標(biāo)內(nèi)部顏色特征與陰影區(qū)域接近而被誤當(dāng)作陰影,降低目標(biāo)的誤檢率。2)提出了一種多目標(biāo)遮擋狀態(tài)下的目標(biāo)跟蹤策略。對(duì)多車輛目標(biāo)跟蹤問(wèn)題,本文采用基于目標(biāo)幾何特征結(jié)合Kalman預(yù)測(cè)器的目標(biāo)跟蹤方法,針對(duì)多目標(biāo)出現(xiàn)遮擋的情況,提出了一種以質(zhì)心跟蹤為主,面積信息為輔,根據(jù)預(yù)測(cè)目標(biāo)與實(shí)際目標(biāo)的個(gè)數(shù)分組討論的跟蹤策略。3)針對(duì)雨雪天氣情況下公路路面狀態(tài)檢測(cè)預(yù)警問(wèn)題,采用樸素貝葉斯分類器對(duì)道路狀態(tài)進(jìn)行分類。首先,提取出圖像中道路的區(qū)域;然后提取道路圖像的熵、能量、對(duì)比度、平均亮度和平均飽和度等參數(shù)作為特征,使用樸素貝葉斯分類器對(duì)道路區(qū)域狀態(tài)分類,并對(duì)濕滑、積雪較多的道路進(jìn)行告警。4)設(shè)計(jì)了高速公路智能視頻監(jiān)控系統(tǒng)。系統(tǒng)可以完成交通視頻中路面背景提取、運(yùn)動(dòng)車輛檢測(cè)與跟蹤、車輛信息統(tǒng)計(jì)、道路雨雪狀態(tài)檢測(cè)等功能,并且可以對(duì)異常交通情況和惡劣道路狀態(tài)進(jìn)行告警處理。
[Abstract]:Expressway is the main line of urban express traffic and has become the backbone of economic development. The safety level of expressway pavement state is greatly affected by natural factors such as rainfall fog snowfall and snowmelt will affect the operation of expressway. Based on the development of intelligent video surveillance system for freeway, this paper studies the technology of moving target recognition, moving vehicle tracking and road rain and snow state detection in expressway surveillance video. And on the basis of the application system design and functional module development. The main work of this paper includes: 1) A new target shadow elimination method is proposed. In order to detect the moving vehicle target accurately, it is necessary to remove the moving shadow first. Aiming at the deficiency of traditional shadow elimination method in HSV color space, a shadow cancellation method combining image region information in HSV color space is proposed in this paper. By using this method, we can avoid being mistaken for shadow because the inner color feature of vehicle is close to the shadow area, and reduce the false detection rate of the target. (2) A target tracking strategy in multi-target occlusion state is proposed. For the multi-vehicle target tracking problem, this paper uses the target tracking method based on the geometric feature of the target and Kalman predictor. In view of the occlusion of the multi-target, a new method is proposed, which is based on the centroid tracking and the area information. According to the tracking strategy. 3) aiming at road pavement condition detection and warning problem in rain and snow weather, naive Bayes classifier is used to classify road state. Firstly, the road region in the image is extracted, and then the parameters such as entropy, energy, contrast, average brightness and average saturation of the road image are extracted as features, and the state of the road region is classified by naive Bayes classifier, and the road is wet and slippery. The highway intelligent video surveillance system is designed. The system can complete the functions of road background extraction, moving vehicle detection and tracking, vehicle information statistics, road rain and snow state detection and so on, and can alarm the abnormal traffic situation and the bad road state.
【學(xué)位授予單位】：北京工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：U495;TP391.41;TN948.6
，

本文編號(hào)：2000984