基于GPU的圖像特征提取加速算法
發(fā)布時(shí)間：2019-03-20 21:16
【摘要】：目前,相對于傳統(tǒng)的文本數(shù)據(jù)來說,圖像和視頻已經(jīng)逐漸成為因特網(wǎng)上傳輸和處理的主要數(shù)據(jù)類型。然而許多針對圖像和視頻的應(yīng)用(如搜索引擎和網(wǎng)絡(luò)信息過濾系統(tǒng))由于圖像檢索算法的限制而無法滿足因特網(wǎng)發(fā)展的需求。目前的圖像檢索算法主要基于全局特征和局部特征兩種�；谌痔卣鞯臋z索算法利用顏色、紋理、形狀或空間關(guān)系等特征來描述圖像或視頻中的一幀,但這些算法只用一個(gè)特征向量進(jìn)行描述,所以速度快但精度不高,無法很好地滿足圖像和視頻檢索的需求�；诰植刻卣鞯臋z索算法用數(shù)百甚至上千個(gè)特征來描述一幅圖像或視頻的一幀,因此具有很高的精度。但由于基于局部特征的圖像特征提取算法計(jì)算比較復(fù)雜,運(yùn)行速度較慢,限制了它的應(yīng)用。因此,加速基于局部特征的圖像特征提取算法是關(guān)注的重點(diǎn)。 最近幾年,隨著半導(dǎo)體技術(shù)的發(fā)展和多核技術(shù)的普及,各種并行硬件逐漸成為應(yīng)用處理的主流。隨著圖像處理單元GPU通用性和可編程性的增強(qiáng),它也成為了其中不可或缺的一個(gè)組成部分。現(xiàn)代GPU不僅僅是一個(gè)單純的圖像處理引擎,更是一個(gè)高度并行的可編程處理器。相比于CPU來說,它有著更為強(qiáng)大的算術(shù)處理能力和更高的存儲器帶寬,這使得它在實(shí)時(shí)處理領(lǐng)域和高性能計(jì)算領(lǐng)域得到了廣泛的應(yīng)用。 本文在GPU上設(shè)計(jì)和實(shí)現(xiàn)了SIFT和SURF的并行加速算法,在我們的實(shí)現(xiàn)中充分利用了GPU的特性來加速我們的實(shí)現(xiàn),包括使用共享內(nèi)存和紋理存儲的使用,盡量減少顯存的分配和釋放次數(shù)等等。GPU通常與CPU協(xié)同工作,傳統(tǒng)的優(yōu)化策略主要關(guān)注GPU上實(shí)現(xiàn)的效率,而忽略了CPU對系統(tǒng)性能的影響。本文在設(shè)計(jì)時(shí)考慮合理分配和利用CPU的資源,進(jìn)一步提高了系統(tǒng)整體性能。本文使用640*480像素大小的圖片測試結(jié)果表明,相對于CPU上的串行版本而言,SIFT的實(shí)現(xiàn)達(dá)到了143.7倍的加速,吞吐量達(dá)到93.39幀/秒,而SURF的實(shí)現(xiàn)達(dá)到了253.2倍的加速,吞吐量達(dá)到346.82幀/秒,很好地滿足了圖像特征提取的實(shí)時(shí)處理的需求�？紤]到SIFT與SURF處理速度的差距,我們建議在圖像匹配的實(shí)時(shí)處理系統(tǒng)中使用SURF算法來進(jìn)行圖像特征的提取。
[Abstract]:At present, compared with the traditional text data, image and video have gradually become the main data types transmitted and processed on the Internet. However, many applications for image and video, such as search engine and network information filtering system, are unable to meet the needs of the development of Internet due to the limitation of image retrieval algorithms. Current image retrieval algorithms are mainly based on global features and local features. Global feature-based retrieval algorithms use features such as color, texture, shape or spatial relationship to describe a frame in an image or video, but these algorithms are described by only one feature vector, so the speed is fast but the precision is not high. It can not meet the needs of image and video retrieval. The local feature-based retrieval algorithm uses hundreds or even thousands of features to describe a frame of an image or video, so it has high accuracy. However, the computation of image feature extraction algorithm based on local feature is complicated and the running speed is slow, which limits its application. Therefore, accelerating the image feature extraction algorithm based on local features is the focus of attention. In recent years, with the development of semiconductor technology and the popularization of multi-core technology, various parallel hardware has gradually become the mainstream of application processing. With the enhancement of generality and programmability of image processing unit (GPU), it has become an indispensable part of it. Modern GPU is not only a simple image processing engine, but also a highly parallel programmable processor. Compared with CPU, it has more powerful arithmetic processing ability and higher memory bandwidth, which makes it widely used in real-time processing and high-performance computing. In this paper, we design and implement a parallel acceleration algorithm for SIFT and SURF on GPU. In our implementation, we make full use of the features of GPU to accelerate our implementation, including the use of shared memory and texture storage. GPUs usually work with CPU. Traditional optimization strategies focus on the efficiency of implementation on GPU and ignore the impact of CPU on system performance. In this paper, the rational allocation and utilization of CPU resources are considered to further improve the overall performance of the system. In this paper, the result of image test with the size of 640 pixels shows that compared with the serial version on CPU, the implementation of SIFT is 143.7 times faster, the throughput is 93.39 frames / second, and the implementation of SURF is 253.2 times faster. The throughput is up to 346.82 frames / sec, which satisfies the real-time processing requirements of image feature extraction. Considering the difference of processing speed between SIFT and SURF, we propose to use SURF algorithm to extract image features in real-time image matching processing system.
【學(xué)位授予單位】：復(fù)旦大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：TP391.41
[Abstract]:At present, compared with the traditional text data, image and video have gradually become the main data types transmitted and processed on the Internet. However, many applications for image and video, such as search engine and network information filtering system, are unable to meet the needs of the development of Internet due to the limitation of image retrieval algorithms. Current image retrieval algorithms are mainly based on global features and local features. Global feature-based retrieval algorithms use features such as color, texture, shape or spatial relationship to describe a frame in an image or video, but these algorithms are described by only one feature vector, so the speed is fast but the precision is not high. It can not meet the needs of image and video retrieval. The local feature-based retrieval algorithm uses hundreds or even thousands of features to describe a frame of an image or video, so it has high accuracy. However, the computation of image feature extraction algorithm based on local feature is complicated and the running speed is slow, which limits its application. Therefore, accelerating the image feature extraction algorithm based on local features is the focus of attention. In recent years, with the development of semiconductor technology and the popularization of multi-core technology, various parallel hardware has gradually become the mainstream of application processing. With the enhancement of generality and programmability of image processing unit (GPU), it has become an indispensable part of it. Modern GPU is not only a simple image processing engine, but also a highly parallel programmable processor. Compared with CPU, it has more powerful arithmetic processing ability and higher memory bandwidth, which makes it widely used in real-time processing and high-performance computing. In this paper, we design and implement a parallel acceleration algorithm for SIFT and SURF on GPU. In our implementation, we make full use of the features of GPU to accelerate our implementation, including the use of shared memory and texture storage. GPUs usually work with CPU. Traditional optimization strategies focus on the efficiency of implementation on GPU and ignore the impact of CPU on system performance. In this paper, the rational allocation and utilization of CPU resources are considered to further improve the overall performance of the system. In this paper, the result of image test with the size of 640 pixels shows that compared with the serial version on CPU, the implementation of SIFT is 143.7 times faster, the throughput is 93.39 frames / second, and the implementation of SURF is 253.2 times faster. The throughput is up to 346.82 frames / sec, which satisfies the real-time processing requirements of image feature extraction. Considering the difference of processing speed between SIFT and SURF, we propose to use SURF algorithm to extract image features in real-time image matching processing system.
【學(xué)位授予單位】：復(fù)旦大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：TP391.41
【相似文獻(xiàn)】
相關(guān)期刊論文 前10條
1 范羚,吳小培,龍飛,張道信,郭曉靜;基于獨(dú)立分量分析的圖像特征提取及去噪[J];計(jì)算機(jī)工程與應(yīng)用;2003年09期
2 林明星,王曉華,管志光,丁鳳華,趙永瑞;基于差分碼的圖像特征提取方法研究[J];儀器儀表學(xué)報(bào);2004年S2期
3 許世軍;楊曉東;任小玲;;光學(xué)圖像特征提取與識別的智能算子發(fā)展研究[J];信息系統(tǒng)工程;2010年10期
4 杜海順;李玉玲;侯彥東;金勇;;一種人臉圖像特征提取的局部和整體間距嵌入方法[J];計(jì)算機(jī)科學(xué);2012年09期
5 孔銳;施澤生;郭立;張國宣;;獨(dú)立分量分析在圖像特征提取中的應(yīng)用[J];模式識別與人工智能;2004年01期
6 周開軍;陽春華;牟學(xué)民;桂衛(wèi)華;;一種基于圖像特征提取的浮選回收率預(yù)測算法[J];高技術(shù)通訊;2009年09期
7 趙英亮;王黎明;韓焱;;基于形態(tài)學(xué)與聚類相結(jié)合的圖像特征提取方法研究[J];彈箭與制導(dǎo)學(xué)報(bào);2010年02期
8 周春光;孫明芳;王u&菁;陳前;劉小華;劉昱昊;;基于稀疏張量的人臉圖像特征提取[J];吉林大學(xué)學(xué)報(bào)(工學(xué)版);2012年06期
9 韓吉衢;孟俊敏;趙俊生;;海洋溢油合成孔徑雷達(dá)圖像特征提取及其關(guān)鍵度分析[J];海洋學(xué)報(bào)(中文版);2013年01期
10 韋v，
本文編號：2444638
本文鏈接：http://sikaile.net/kejilunwen/sousuoyinqinglunwen/2444638.html
最近更新
教材專著
