本文選題：LDA主題模型 + K-means++�。� 參考：《中北大學(xué)》2017年碩士論文

【摘要】：圖像分類(lèi)作為計(jì)算機(jī)視覺(jué)領(lǐng)域研究的方向之一,是其他圖像應(yīng)用領(lǐng)域的基礎(chǔ)。此類(lèi)問(wèn)題的一個(gè)重要解決途徑就是圖像場(chǎng)景分類(lèi)技術(shù)。本文主要針對(duì)現(xiàn)有的基于LDA主題模型圖像場(chǎng)景分類(lèi)技術(shù)中存在的一些問(wèn)題,提出新的改進(jìn)的方法,以提高LDA主題模型對(duì)圖像場(chǎng)景分類(lèi)的準(zhǔn)確率和執(zhí)行效率。潛在狄利克雷分布(Latent Dirichlet Allocation,LDA)主題模型是當(dāng)前廣泛使用的一種圖像處理方法,它將圖像的底層局部特征抽象成為視覺(jué)單詞,生成視覺(jué)詞典,統(tǒng)計(jì)視覺(jué)單詞出現(xiàn)的頻率進(jìn)而建立中層語(yǔ)義表示模型,對(duì)圖像進(jìn)行表示。之后通過(guò)分類(lèi)器自動(dòng)標(biāo)記圖像場(chǎng)景標(biāo)簽,實(shí)現(xiàn)圖像的自動(dòng)分類(lèi)。針對(duì)傳統(tǒng)模型在進(jìn)行圖像場(chǎng)景識(shí)別時(shí)存在的問(wèn)題,本文進(jìn)行了如下研究:1.針對(duì)傳統(tǒng)模型在進(jìn)行圖像場(chǎng)景識(shí)別時(shí)使用的聚類(lèi)方法效率較低的問(wèn)題,采用KMeans++聚類(lèi)算法生成視覺(jué)單詞。2.傳統(tǒng)方法表示圖像時(shí)未考慮單詞的權(quán)重問(wèn)題,使得學(xué)習(xí)得到的主題分布傾向高頻詞,針對(duì)視覺(jué)單詞出現(xiàn)的冪律分布問(wèn)題,本文使用加權(quán)統(tǒng)計(jì)直方圖進(jìn)行圖像表示。3.在進(jìn)行圖像場(chǎng)景識(shí)別時(shí)不能有效利用圖像主要特征的問(wèn)題,引入特征函數(shù),在圖像場(chǎng)景識(shí)別模型的方法中加強(qiáng)重要特征在分類(lèi)識(shí)別中的作用,提出有特征函數(shù)的潛在狄利克雷分布(Featured Latent Dirichlet Allocation,FLDA)主題模型,提高圖像場(chǎng)景的分類(lèi)和識(shí)別效率。4.LDA模型中的參數(shù)很難直接估計(jì),針對(duì)這個(gè)問(wèn)題,提出了一種改進(jìn)的變分推理方法,即快速變分推理(Fast Variational Inference,FVI),減少模型中參數(shù)的迭代次數(shù),減少計(jì)算成本,提高模型的執(zhí)行效率。通過(guò)對(duì)不同數(shù)據(jù)集上的多次實(shí)驗(yàn)結(jié)果進(jìn)行分析可知,本文提出的FLDA模型和快速變分推理算法,能夠有效的提高基于主題模型的圖像場(chǎng)景分類(lèi)的準(zhǔn)確率和執(zhí)行效率,并且具有一定的通用性和穩(wěn)定性。
[Abstract]:As one of the research directions in the field of computer vision, image classification is the basis of other image application fields. One of the most important ways to solve this problem is image scene classification. In order to improve the accuracy and efficiency of image scene classification based on LDA topic model, this paper proposes a new and improved method to solve some problems existing in the existing image scene classification technology based on LDA topic model. The topic model of latent Dirichlet allocation LDA (LDA) is a widely used image processing method, which abstracts the underlying local features of images into visual words and generates visual dictionaries. Then the middle level semantic representation model is established to represent the image by counting the frequency of visual words. Then the image scene label is automatically marked by classifier to realize the automatic image classification. Aiming at the problems of traditional model in image scene recognition, this paper does the following research: 1: 1. Aiming at the low efficiency of the traditional clustering method used in image scene recognition, KMeans clustering algorithm is used to generate visual words. 2. The traditional method does not consider the weight of words, which makes the topic distribution tend to high frequency words. In view of the power law distribution of visual words, the weighted statistical histogram is used to represent the image. 3. In the process of image scene recognition, the main features of the image can not be effectively utilized. The feature function is introduced to strengthen the important features in the classification and recognition of the image scene recognition model. In this paper, a Featured Latent Dirichlet allocation / FLDA subject model is proposed, which improves the classification and recognition efficiency of images. 4. The parameters in LDA model are difficult to estimate directly. In order to solve this problem, an improved variational reasoning method is proposed. Fast variational inference (FVI) reduces the number of iterations of parameters in the model, reduces the computational cost and improves the execution efficiency of the model. Through the analysis of many experiments on different data sets, we can see that the FLDA model and the fast variational reasoning algorithm proposed in this paper can effectively improve the accuracy and efficiency of image scene classification based on topic model. And has certain universality and stability.
【學(xué)位授予單位】：中北大學(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10條

1 顧廣華;韓晰瑛;陳春霞;趙耀;;圖像場(chǎng)景語(yǔ)義分類(lèi)研究進(jìn)展綜述[J];系統(tǒng)工程與電子技術(shù);2016年04期

2 韓霞;黃德根;;基于半監(jiān)督隱馬爾科夫模型的漢語(yǔ)詞性標(biāo)注研究[J];小型微型計(jì)算機(jī)系統(tǒng);2015年12期

3 李學(xué)龍;史建華;董永生;陶大程;;場(chǎng)景圖像分類(lèi)技術(shù)綜述[J];中國(guó)科學(xué):信息科學(xué);2015年07期

4 田仙仙;鮑泓;徐成;;一種改進(jìn)HOG特征的行人檢測(cè)算法[J];計(jì)算機(jī)科學(xué);2014年09期

5 趙仲秋;季海峰;高雋;胡東輝;吳信東;;基于稀疏編碼多尺度空間潛在語(yǔ)義分析的圖像分類(lèi)[J];計(jì)算機(jī)學(xué)報(bào);2014年06期

6 張瑞杰;李弼程;魏福山;;基于多尺度上下文語(yǔ)義信息的圖像場(chǎng)景分類(lèi)算法[J];電子學(xué)報(bào);2014年04期

7 趙理君;唐娉;霍連志;鄭柯;;圖像場(chǎng)景分類(lèi)中視覺(jué)詞包模型方法綜述[J];中國(guó)圖象圖形學(xué)報(bào);2014年03期

8 華媛蕾;劉萬(wàn)軍;;改進(jìn)混合高斯模型的運(yùn)動(dòng)目標(biāo)檢測(cè)算法[J];計(jì)算機(jī)應(yīng)用;2014年02期

9 周書(shū)仁;殷建平;;基于Haar特性的LBP紋理特征[J];軟件學(xué)報(bào);2013年08期

10 都云程;任紹美;王濤;呂學(xué)強(qiáng);;基于空間金字塔的鏡頭檢測(cè)[J];計(jì)算機(jī)工程與應(yīng)用;2013年11期

，

本文編號(hào)：2089856