基于多特征的行人快速檢測(cè)方法研究
發(fā)布時(shí)間：2018-10-29 09:30
【摘要】：當(dāng)前計(jì)算機(jī)視覺(jué)領(lǐng)域研究的熱點(diǎn)之一就是人臉識(shí)別和行人檢測(cè),這一技術(shù)已經(jīng)被廣泛的應(yīng)用在很多領(lǐng)域,比如智能電話、智能交通、無(wú)人駕駛等。由于算法的精度和速度等原因,很難應(yīng)用到實(shí)時(shí)系統(tǒng)中。傳統(tǒng)方法中為了提高精度需要計(jì)算圖像特征金字塔,而構(gòu)建特征金字塔需要花費(fèi)大量時(shí)間。本文正是從這一問(wèn)題出發(fā),通過(guò)估算多尺度分類(lèi)器的方法,減少構(gòu)建特征金字塔時(shí)間從而提高了行人檢測(cè)的速度。本文主要的研究?jī)?nèi)容如下:(1)本文提出了一種新特征BPG。BPG特征實(shí)質(zhì)是HOG的一種變形,它既保留了原有的梯度和方向信息,也具有局部區(qū)域信息。BPG特征把梯度方向分成8個(gè)方向,在可變大小區(qū)域內(nèi)的不同梯度方向上累加梯度值,然后與均值作比較,進(jìn)行二值編碼,最后生成十進(jìn)制數(shù)。實(shí)驗(yàn)結(jié)果顯示,新特征在行人檢測(cè)方面有更強(qiáng)的識(shí)別能力。(2)通過(guò)實(shí)驗(yàn)對(duì)比和特征間的相關(guān)性挑選出四個(gè)特征作為特征池,四個(gè)特征分別是BPG特征、LBP特征、梯度特征值和下梯度方向特征。根據(jù)特征本身的特點(diǎn)分析了特征間的互補(bǔ)性。用這四個(gè)特征融合對(duì)單一樣本的檢測(cè)正確率大約為97%。(3)分類(lèi)器設(shè)計(jì)主要使用的是Adaboost算法。通過(guò)級(jí)聯(lián)形式,使每級(jí)強(qiáng)分類(lèi)器有不同數(shù)量的弱分類(lèi)器和閾值,使分類(lèi)器的識(shí)別能力逐級(jí)加強(qiáng)。這樣可以將容易識(shí)別出的負(fù)樣本在第一級(jí)或是前幾級(jí)分類(lèi)器就排除掉,難以識(shí)別的由后面識(shí)別能力強(qiáng)的分類(lèi)器識(shí)別,既提高了分類(lèi)器的精度又減少了檢測(cè)窗口的數(shù)量。(4)提出了一種估計(jì)分類(lèi)器的方法,通過(guò)估計(jì)臨近放縮層的分類(lèi)器以代替圖像的縮放過(guò)程,大幅度減少計(jì)算特征金子塔所耗費(fèi)的時(shí)間,實(shí)驗(yàn)表明估計(jì)分類(lèi)器算法雖然會(huì)使檢測(cè)的精度下降1-3%,但檢測(cè)的速度卻提升了 2倍多。
[Abstract]:Face recognition and pedestrian detection are one of the hot topics in the field of computer vision. This technology has been widely used in many fields, such as smart phones, intelligent transportation, driverless and so on. Because of the accuracy and speed of the algorithm, it is difficult to be applied to real-time system. In order to improve the accuracy of traditional methods, we need to calculate the image feature pyramid, but it takes a lot of time to construct the feature pyramid. From this point of view, the method of estimating multi-scale classifiers is used to reduce the time of constructing feature pyramids and improve the speed of pedestrian detection. The main contents of this paper are as follows: (1) in this paper, we propose a new feature, BPG.BPG feature, which is essentially a kind of deformation of HOG, which not only preserves the original gradient and direction information. The BPG feature divides the gradient direction into eight directions and accumulates the gradient value in the different gradient directions in the variable size region. Then compared with the mean value the binary coding is carried out and finally the decimal number is generated. The experimental results show that the new features have stronger recognition ability in pedestrian detection. (2) four features are selected as feature pool through experimental comparison and correlation among the features. The four features are BPG features and LBP features. Gradient eigenvalues and lower gradient directional features. The complementarities between features are analyzed according to the characteristics of the features themselves. The accuracy of detecting a single sample with these four features fusion is about 97. (3) the classifier is designed mainly using Adaboost algorithm. By cascading, each strong classifier has a different number of weak classifiers and threshold values, and the recognition ability of the classifier is enhanced step by step. In this way, the negative samples that are easy to identify can be excluded at the first or first stages of the classifier, and those that are difficult to recognize are identified by the classifier with strong recognition ability behind them. It not only improves the accuracy of classifiers but also reduces the number of detection windows. (4) A method of estimating classifiers is proposed to replace the zooming process of images by estimating the classifiers near the scaling layer. The experimental results show that the estimation classifier algorithm can reduce the accuracy of detection by 1-3 and increase the speed of detection by more than 2 times.
【學(xué)位授予單位】：內(nèi)蒙古大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP391.41
本文編號(hào)：2297311
[Abstract]:Face recognition and pedestrian detection are one of the hot topics in the field of computer vision. This technology has been widely used in many fields, such as smart phones, intelligent transportation, driverless and so on. Because of the accuracy and speed of the algorithm, it is difficult to be applied to real-time system. In order to improve the accuracy of traditional methods, we need to calculate the image feature pyramid, but it takes a lot of time to construct the feature pyramid. From this point of view, the method of estimating multi-scale classifiers is used to reduce the time of constructing feature pyramids and improve the speed of pedestrian detection. The main contents of this paper are as follows: (1) in this paper, we propose a new feature, BPG.BPG feature, which is essentially a kind of deformation of HOG, which not only preserves the original gradient and direction information. The BPG feature divides the gradient direction into eight directions and accumulates the gradient value in the different gradient directions in the variable size region. Then compared with the mean value the binary coding is carried out and finally the decimal number is generated. The experimental results show that the new features have stronger recognition ability in pedestrian detection. (2) four features are selected as feature pool through experimental comparison and correlation among the features. The four features are BPG features and LBP features. Gradient eigenvalues and lower gradient directional features. The complementarities between features are analyzed according to the characteristics of the features themselves. The accuracy of detecting a single sample with these four features fusion is about 97. (3) the classifier is designed mainly using Adaboost algorithm. By cascading, each strong classifier has a different number of weak classifiers and threshold values, and the recognition ability of the classifier is enhanced step by step. In this way, the negative samples that are easy to identify can be excluded at the first or first stages of the classifier, and those that are difficult to recognize are identified by the classifier with strong recognition ability behind them. It not only improves the accuracy of classifiers but also reduces the number of detection windows. (4) A method of estimating classifiers is proposed to replace the zooming process of images by estimating the classifiers near the scaling layer. The experimental results show that the estimation classifier algorithm can reduce the accuracy of detection by 1-3 and increase the speed of detection by more than 2 times.
【學(xué)位授予單位】：內(nèi)蒙古大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP391.41
【參考文獻(xiàn)】
相關(guān)期刊論文 前2條
1 王志國(guó);王貴錦;施陳博;苗權(quán);林行剛;;積分圖像的快速GPU計(jì)算[J];計(jì)算機(jī)應(yīng)用研究;2011年10期
2 周欣,黃席樾,樊友平,劉濤;汽車(chē)智能輔助駕駛系統(tǒng)中的單目視覺(jué)導(dǎo)航技術(shù)[J];機(jī)器人;2003年04期
相關(guān)碩士學(xué)位論文 前1條
1 裴文娟;基于梯度特征和紋理特征的行人檢測(cè)[D];安徽大學(xué);2015年
，本文編號(hào)：2297311
本文鏈接：http://www.lk138.cn/shoufeilunwen/xixikjs/2297311.html
最近更新
教材專著
