《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 模擬設(shè)計(jì) > 設(shè)計(jì)應(yīng)用 > 公路視頻實(shí)時(shí)車輛檢測分類與車流量統(tǒng)計(jì)算法
公路視頻實(shí)時(shí)車輛檢測分類與車流量統(tǒng)計(jì)算法
2020年信息技術(shù)與網(wǎng)絡(luò)安全第3期
查偉偉,白天
(中國科學(xué)技術(shù)大學(xué) 軟件學(xué)院,安徽 合肥 230000)
摘要: 公路視頻實(shí)時(shí)車輛檢測分類與車流量統(tǒng)計(jì)是計(jì)算機(jī)視覺領(lǐng)域的一個(gè)經(jīng)典問題。傳統(tǒng)設(shè)置檢測帶法,易漏檢復(fù)檢,自動(dòng)化性不好。基于深度網(wǎng)絡(luò)的one-stage算法實(shí)時(shí)性好,但是經(jīng)常會(huì)把變化的背景、運(yùn)動(dòng)的非車輛物體納入其中,同時(shí)對(duì)光照變化敏感,夜間分類效果不好。因此,提出采用one-stage做目標(biāo)檢測,并不直接獲取分類結(jié)果,而是根據(jù)標(biāo)注框?qū)⑽矬w切割出來,去除背景,提升抗背景擾動(dòng)性能和分類效果;再送入一個(gè)經(jīng)過遷移學(xué)習(xí)的淺層神經(jīng)網(wǎng)絡(luò);將分類輸出和目標(biāo)檢測網(wǎng)絡(luò)的位置輸出合并送入一個(gè)全圖匹配算法,進(jìn)行車流量統(tǒng)計(jì)。該方法在保障實(shí)時(shí)性的同時(shí)降低了漏檢和復(fù)檢率。
中圖分類號(hào):TP181
文獻(xiàn)標(biāo)識(shí)碼:A
DOI: 10.19358/j.issn.2096-5133.2020.03.012
引用格式:查偉偉,白天.公路視頻實(shí)時(shí)車輛檢測分類與車流量統(tǒng)計(jì)算法[J].信息技術(shù)與網(wǎng)絡(luò)安全,2020,39(3):6267,72.
Highway video real-time vehicle detection classification and traffic flow statistics algorithm
Zha Weiwei, Bai Tian
(Department of Software Engineering,University of Science and Technology of China,Hefei 230000,China)
Abstract: Realtime vehicle detection,classification and traffic statistics based on road video are classic problems in the field of computer vision.The traditional method of setting the detection belt is prone to missed inspection and reinspection,so the automation performance is not good.The realtime performance of onestage algorithm based on deep network can be guaranteed,but the changing background,moving nonvehicle objects are often included,and the change of illumination is sensitive at the same time,so the classification at night is not good.Therefore,an algorithm is proposed to perform target detection by onestage,and the classification result is not directly obtained. Instead, it cuts out the object according to the bounding box,removes the background, and improves resistance to background disturbances and classification accuracy.Then it is sent to a transfer learning shallow neural network. The classified output and the position output of the target detection network are combined and sent to a full map matching algorithm for traffic flow statistics.While ensuring realtime performance,the rate of missed inspections and reinspections is reduced.
Key words : convolutional neural network;target detection and classification;realtime traffic statistics;YOLOv3 network

0     引言

公路視頻的車輛分類與車流量統(tǒng)計(jì)是運(yùn)動(dòng)物體目標(biāo)檢測識(shí)別與跟蹤問題,可以通過傳統(tǒng)圖像方法和現(xiàn)代深度網(wǎng)絡(luò)實(shí)現(xiàn)。傳統(tǒng)圖像方法由于計(jì)算量較小,因此實(shí)時(shí)性相對(duì)較高?,F(xiàn)代深度網(wǎng)絡(luò)在背景分割、目標(biāo)分類的準(zhǔn)確度上有著壓倒性的優(yōu)勢(shì)。




本文詳細(xì)內(nèi)容請(qǐng)下載:http://ihrv.cn/resource/share/2000003187





作者信息:

查偉偉,白天

(中國科學(xué)技術(shù)大學(xué) 軟件學(xué)院,安徽 合肥 230000)



此內(nèi)容為AET網(wǎng)站原創(chuàng),未經(jīng)授權(quán)禁止轉(zhuǎn)載。