【摘要】：隨著視頻分析、處理和傳輸技術(shù)要求的不斷提高,視頻監(jiān)控系統(tǒng)已由原來的模擬方式向數(shù)字化、智能化方向改進(jìn),但卻導(dǎo)致了視頻監(jiān)控?cái)?shù)據(jù)的急劇增加,使得現(xiàn)有的視頻編碼標(biāo)準(zhǔn)已不能滿足新型視頻應(yīng)用的發(fā)展要求。新一代高效視頻編碼標(biāo)準(zhǔn)H.265/HEVC(High Efficiency Vedio Coding)提供了多項(xiàng)新的技術(shù),隨著自身不斷完善的同時將逐漸取代H.264/AVC,并逐步應(yīng)用于各類視頻編碼業(yè)務(wù)中。HEVC采用的新技術(shù)使得編碼效率相比于H.264提高了一倍,但這些新技術(shù)也使得編碼計(jì)算復(fù)雜度成倍增加。近年來,如何采用有效的方法減小HEVC的編碼計(jì)算復(fù)雜度一直是HEVC視頻編碼研究工作的重中之重。為了提高編碼效率,本文將智能視頻監(jiān)控技術(shù)和HEVC編碼標(biāo)準(zhǔn)相結(jié)合,從HEVC編碼框架的關(guān)鍵模塊入手,深入學(xué)習(xí)各關(guān)鍵編碼模塊的技術(shù)原理,通過詳細(xì)研究編碼單元塊(Coding Unit,CU)的劃分算法和熵編碼技術(shù)原理,提出了基于高關(guān)注度區(qū)域?qū)U塊劃分和熵編碼技術(shù)的改進(jìn)算法。首先為本文的改進(jìn)算法做前期工作,提取視頻序列中的高關(guān)注區(qū)域,也稱作感興趣區(qū)域即監(jiān)控視頻中的運(yùn)動目標(biāo)區(qū)域,感興趣區(qū)域主要利用Kinect和立體視頻的深度信息和混合高斯模型法提取視頻序列中的背景信息結(jié)合深度信息定義和背景差分法得到;其次將提取的感興趣區(qū)域關(guān)聯(lián)到視頻的編解碼過程中,針對已獲取的感興趣區(qū)域,對HEVC中CU塊劃分提出提前終止策略,減少不必要的率失真代價(jià)計(jì)算,加快編碼效率,該方法在保證視頻編碼質(zhì)量損失較小的前提下,平均編碼時長縮減54.35%,顯著提升了編碼效率;最后考慮到CU塊劃分對熵編碼的影響,利用感興趣區(qū)域?qū)υ悸适д嫠惴ㄟM(jìn)行改進(jìn),在保證視頻圖像質(zhì)量和編碼復(fù)雜度相對不變的情況下,減少了51.6%的比特率,明顯改善了HEVC原有算法的壓縮性能。
[Abstract]:With the continuous improvement of video analysis, processing and transmission technology, video surveillance system has been improved from the original analog mode to the direction of digitalization and intelligence, but it has led to a sharp increase of video surveillance data. So that the existing video coding standards can no longer meet the requirements of the development of new video applications. The new generation of efficient video coding standard H.265/HEVC (High Efficiency Vedio Coding) offers a number of new technologies that will gradually replace H.264 AVC as it continues to be perfected. The new technologies adopted by HEVC make the coding efficiency more than double that of H.264, but these new technologies also double the complexity of coding. In recent years, how to use effective methods to reduce the computational complexity of HEVC coding has been the most important task in the research of HEVC video coding. In order to improve the coding efficiency, this paper combines the intelligent video surveillance technology with the HEVC coding standard, starts with the key modules of the HEVC coding framework, deeply studies the technical principles of the key coding modules, and studies the coding unit block (Coding Unit, in detail. Based on the partition algorithm of CU and the principle of entropy coding, an improved algorithm for block partition and entropy coding of CU based on high attention region is proposed. First of all, we do the preliminary work for the improved algorithm in this paper, and extract the high attention region in the video sequence, which is also called the region of interest, that is, the moving target area in the surveillance video. The region of interest mainly uses the depth information of Kinect and stereo video and the mixed Gao Si model method to extract the background information from the video sequence combined with the definition of depth information and the background difference method. Secondly, the extracted region of interest is associated with the video encoding and decoding process. Aiming at the acquired region of interest, an early termination strategy is proposed for the partition of CU blocks in HEVC, which reduces unnecessary rate-distortion cost calculation and speeds up coding efficiency. On the premise that the video coding quality loss is small, the average coding time is reduced by 54.35%, and the coding efficiency is significantly improved. Finally, considering the effect of CU block partition on entropy coding, the original rate-distortion algorithm is improved by using the region of interest, and the bit rate is reduced by 51.6% while the video image quality and coding complexity are relatively unchanged. The compression performance of the original HEVC algorithm is improved obviously.
【學(xué)位授予單位】：中原工學(xué)院
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TN919.81

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 詹曉濤;;HEVC幀內(nèi)預(yù)測編碼快速判決算法分析[J];中國有線電視;2015年07期

2 王璐;;智能視頻監(jiān)控技術(shù)應(yīng)用[J];中國科技信息;2015年05期

3 黃凱奇;陳曉棠;康運(yùn)鋒;譚鐵牛;;智能視頻監(jiān)控技術(shù)綜述[J];計(jì)算機(jī)學(xué)報(bào);2015年06期

4 許東旭;林其偉;;高效率視頻編碼幀內(nèi)快速算法[J];計(jì)算機(jī)應(yīng)用;2014年08期

5 于成業(yè);何小海;滕奇志;鐘國韻;李元;;基于非規(guī)則PU模式的HEVC視頻編碼方法[J];電視技術(shù);2014年09期

6 阮若林;胡瑞敏;;數(shù)字視頻壓縮編碼技術(shù)標(biāo)準(zhǔn)現(xiàn)狀與展望[J];電視技術(shù);2014年03期

7 戰(zhàn)蔭偉;張昊;;基于Kinect傳感器的人體行為分析算法[J];傳感器與微系統(tǒng);2015年01期

8 鐘國韻;湯彬;徐洪珍;吳有用;;時空域相關(guān)性的快速HEVC幀間模式判決方法[J];東華理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2013年04期

9 徐飛;張永飛;李哲;;HEVC熵編碼算法優(yōu)化[J];計(jì)算機(jī)與現(xiàn)代化;2013年12期

10 沈燕飛;李錦濤;朱珍民;張勇東;;高效視頻編碼[J];計(jì)算機(jī)學(xué)報(bào);2013年11期

，

本文編號：2472600