Distribution Entropy Boosted VLAD for Image Retrieval
Several recent works have shown that aggregating local descriptors to generate global image representation results in great efficiency for retrieval and classification tasks. The most popular method following this approach is VLAD (Vector of Locally Aggregated Descriptors). We present a novel image...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2016-08-01
|
Series: | Entropy |
Subjects: | |
Online Access: | http://www.mdpi.com/1099-4300/18/8/311 |
id |
doaj-f0ebdfd3a44345258240f2e379c65951 |
---|---|
record_format |
Article |
spelling |
doaj-f0ebdfd3a44345258240f2e379c659512020-11-24T21:23:18ZengMDPI AGEntropy1099-43002016-08-0118831110.3390/e18080311e18080311Distribution Entropy Boosted VLAD for Image RetrievalQiuzhan Zhou0Cheng Wang1Pingping Liu2Qingliang Li3Yeran Wang4Shuozhang Chen5State Key Laboratory of Automotive Simulation and Control, Jilin University, Changchun 130022, ChinaCollege of Communication Engineering, Jilin University, Changchun 130012, ChinaCollege of Computer Science and Technology, Jilin University, Changchun 130012, ChinaCollege of Computer Science and Technology, Jilin University, Changchun 130012, ChinaCollege of Computer Science and Technology, Jilin University, Changchun 130012, ChinaCollege of Communication Engineering, Jilin University, Changchun 130012, ChinaSeveral recent works have shown that aggregating local descriptors to generate global image representation results in great efficiency for retrieval and classification tasks. The most popular method following this approach is VLAD (Vector of Locally Aggregated Descriptors). We present a novel image presentation called Distribution Entropy Boosted VLAD (EVLAD), which extends the original vector of locally aggregated descriptors. The original VLAD adopts only residuals to depict the distribution information of every visual word and neglects other statistical clues, so its discriminative power is limited. To address this issue, this paper proposes the use of the distribution entropy of each cluster as supplementary information to enhance the search accuracy. To fuse two feature sources organically, two fusion methods after a new normalization stage meeting power law are also investigated, which generate identically sized and double-sized vectors as the original VLAD. We validate our approach in image retrieval and image classification experiments. Experimental results demonstrate the effectiveness of our algorithm.http://www.mdpi.com/1099-4300/18/8/311image retrievalVLADdistribution entropyquantization errornormalization |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Qiuzhan Zhou Cheng Wang Pingping Liu Qingliang Li Yeran Wang Shuozhang Chen |
spellingShingle |
Qiuzhan Zhou Cheng Wang Pingping Liu Qingliang Li Yeran Wang Shuozhang Chen Distribution Entropy Boosted VLAD for Image Retrieval Entropy image retrieval VLAD distribution entropy quantization error normalization |
author_facet |
Qiuzhan Zhou Cheng Wang Pingping Liu Qingliang Li Yeran Wang Shuozhang Chen |
author_sort |
Qiuzhan Zhou |
title |
Distribution Entropy Boosted VLAD for Image Retrieval |
title_short |
Distribution Entropy Boosted VLAD for Image Retrieval |
title_full |
Distribution Entropy Boosted VLAD for Image Retrieval |
title_fullStr |
Distribution Entropy Boosted VLAD for Image Retrieval |
title_full_unstemmed |
Distribution Entropy Boosted VLAD for Image Retrieval |
title_sort |
distribution entropy boosted vlad for image retrieval |
publisher |
MDPI AG |
series |
Entropy |
issn |
1099-4300 |
publishDate |
2016-08-01 |
description |
Several recent works have shown that aggregating local descriptors to generate global image representation results in great efficiency for retrieval and classification tasks. The most popular method following this approach is VLAD (Vector of Locally Aggregated Descriptors). We present a novel image presentation called Distribution Entropy Boosted VLAD (EVLAD), which extends the original vector of locally aggregated descriptors. The original VLAD adopts only residuals to depict the distribution information of every visual word and neglects other statistical clues, so its discriminative power is limited. To address this issue, this paper proposes the use of the distribution entropy of each cluster as supplementary information to enhance the search accuracy. To fuse two feature sources organically, two fusion methods after a new normalization stage meeting power law are also investigated, which generate identically sized and double-sized vectors as the original VLAD. We validate our approach in image retrieval and image classification experiments. Experimental results demonstrate the effectiveness of our algorithm. |
topic |
image retrieval VLAD distribution entropy quantization error normalization |
url |
http://www.mdpi.com/1099-4300/18/8/311 |
work_keys_str_mv |
AT qiuzhanzhou distributionentropyboostedvladforimageretrieval AT chengwang distributionentropyboostedvladforimageretrieval AT pingpingliu distributionentropyboostedvladforimageretrieval AT qingliangli distributionentropyboostedvladforimageretrieval AT yeranwang distributionentropyboostedvladforimageretrieval AT shuozhangchen distributionentropyboostedvladforimageretrieval |
_version_ |
1725992306743443456 |