An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm

Ali Ridho Barakbah, Yasushi Kiyoki

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

This paper presents an image search system with an emotion-oriented context recognition mechanism. Our motivation implementing an emotional context is to express user's impressions for retrieval process in the image search system. This emotional context recognizes the most important features by connecting the user's impressions to the image queries. The Mathematical Model of Meaning (MMM: [2], [4] and [5]) is applied for recognizing a series of emotional contexts for retrieving the most highly correlated impressions to the context. These impressions are then projected to a color impression metric to obtain the most significant colors for subspace feature selection. After applying subspace feature selection, the system then clusters the subspace color features of the image dataset using our proposed Pillar-Kmeans algorithm. Pillar algorithm is an algorithm to optimize the initial centroids for K-means clustering. This algorithm is very robust and superior for initial centroids optimization for K-means by positioning all centroids far separately among them in the data distribution. It is inspiring that by distributing the pillars as far as possible from each other within the pressure distribution of a roof, the pillars can withstand the roof's pressure and stabilize a house or building. It considers the pillars which should be located as far as possible from each other to withstand against the pressure distribution of a roof, as number of centroids among the gravity weight of data distribution in the vector space. Therefore, this algorithm designates positions of initial centroids in the farthest accumulated distance between them in the data distribution. The cluster based similarity measurement also involves a semantic filtering mechanism. This mechanism filters out the unimportant image data items to the context in order to speed up the computational execution for image search process. The system then clusters the image dataset using our Pillar-Kmeans algorithm. The centroids of clustering results are used for calculating the similarity measurements to the image query. We perform our proposed system for experimental purpose with the Ukiyo-e image dataset from Tokyo Metropolitan Library for representing the Japanese cultural image collections.

Original languageEnglish
Title of host publicationFrontiers in Artificial Intelligence and Applications
Pages117-136
Number of pages20
Volume225
DOIs
Publication statusPublished - 2011

Publication series

NameFrontiers in Artificial Intelligence and Applications
Volume225
ISSN (Print)09226389

Fingerprint

Roofs
Color
Pressure distribution
Feature extraction
Vector spaces
Gravitation
Semantics
Mathematical models

Keywords

  • cluster based similarity
  • emotional context
  • Image search
  • multi-query images
  • subspace feature selection

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Barakbah, A. R., & Kiyoki, Y. (2011). An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm. In Frontiers in Artificial Intelligence and Applications (Vol. 225, pp. 117-136). (Frontiers in Artificial Intelligence and Applications; Vol. 225). https://doi.org/10.3233/978-1-60750-690-4-117

An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm. / Barakbah, Ali Ridho; Kiyoki, Yasushi.

Frontiers in Artificial Intelligence and Applications. Vol. 225 2011. p. 117-136 (Frontiers in Artificial Intelligence and Applications; Vol. 225).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Barakbah, AR & Kiyoki, Y 2011, An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm. in Frontiers in Artificial Intelligence and Applications. vol. 225, Frontiers in Artificial Intelligence and Applications, vol. 225, pp. 117-136. https://doi.org/10.3233/978-1-60750-690-4-117
Barakbah AR, Kiyoki Y. An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm. In Frontiers in Artificial Intelligence and Applications. Vol. 225. 2011. p. 117-136. (Frontiers in Artificial Intelligence and Applications). https://doi.org/10.3233/978-1-60750-690-4-117
Barakbah, Ali Ridho ; Kiyoki, Yasushi. / An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm. Frontiers in Artificial Intelligence and Applications. Vol. 225 2011. pp. 117-136 (Frontiers in Artificial Intelligence and Applications).
@inproceedings{bf9da5cf5e32491fa2e278d126762f29,
title = "An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm",
abstract = "This paper presents an image search system with an emotion-oriented context recognition mechanism. Our motivation implementing an emotional context is to express user's impressions for retrieval process in the image search system. This emotional context recognizes the most important features by connecting the user's impressions to the image queries. The Mathematical Model of Meaning (MMM: [2], [4] and [5]) is applied for recognizing a series of emotional contexts for retrieving the most highly correlated impressions to the context. These impressions are then projected to a color impression metric to obtain the most significant colors for subspace feature selection. After applying subspace feature selection, the system then clusters the subspace color features of the image dataset using our proposed Pillar-Kmeans algorithm. Pillar algorithm is an algorithm to optimize the initial centroids for K-means clustering. This algorithm is very robust and superior for initial centroids optimization for K-means by positioning all centroids far separately among them in the data distribution. It is inspiring that by distributing the pillars as far as possible from each other within the pressure distribution of a roof, the pillars can withstand the roof's pressure and stabilize a house or building. It considers the pillars which should be located as far as possible from each other to withstand against the pressure distribution of a roof, as number of centroids among the gravity weight of data distribution in the vector space. Therefore, this algorithm designates positions of initial centroids in the farthest accumulated distance between them in the data distribution. The cluster based similarity measurement also involves a semantic filtering mechanism. This mechanism filters out the unimportant image data items to the context in order to speed up the computational execution for image search process. The system then clusters the image dataset using our Pillar-Kmeans algorithm. The centroids of clustering results are used for calculating the similarity measurements to the image query. We perform our proposed system for experimental purpose with the Ukiyo-e image dataset from Tokyo Metropolitan Library for representing the Japanese cultural image collections.",
keywords = "cluster based similarity, emotional context, Image search, multi-query images, subspace feature selection",
author = "Barakbah, {Ali Ridho} and Yasushi Kiyoki",
year = "2011",
doi = "10.3233/978-1-60750-690-4-117",
language = "English",
isbn = "9781607506898",
volume = "225",
series = "Frontiers in Artificial Intelligence and Applications",
pages = "117--136",
booktitle = "Frontiers in Artificial Intelligence and Applications",

}

TY - GEN

T1 - An emotion-oriented image search system with cluster based similarity measurement using pillar-kmeans algorithm

AU - Barakbah, Ali Ridho

AU - Kiyoki, Yasushi

PY - 2011

Y1 - 2011

N2 - This paper presents an image search system with an emotion-oriented context recognition mechanism. Our motivation implementing an emotional context is to express user's impressions for retrieval process in the image search system. This emotional context recognizes the most important features by connecting the user's impressions to the image queries. The Mathematical Model of Meaning (MMM: [2], [4] and [5]) is applied for recognizing a series of emotional contexts for retrieving the most highly correlated impressions to the context. These impressions are then projected to a color impression metric to obtain the most significant colors for subspace feature selection. After applying subspace feature selection, the system then clusters the subspace color features of the image dataset using our proposed Pillar-Kmeans algorithm. Pillar algorithm is an algorithm to optimize the initial centroids for K-means clustering. This algorithm is very robust and superior for initial centroids optimization for K-means by positioning all centroids far separately among them in the data distribution. It is inspiring that by distributing the pillars as far as possible from each other within the pressure distribution of a roof, the pillars can withstand the roof's pressure and stabilize a house or building. It considers the pillars which should be located as far as possible from each other to withstand against the pressure distribution of a roof, as number of centroids among the gravity weight of data distribution in the vector space. Therefore, this algorithm designates positions of initial centroids in the farthest accumulated distance between them in the data distribution. The cluster based similarity measurement also involves a semantic filtering mechanism. This mechanism filters out the unimportant image data items to the context in order to speed up the computational execution for image search process. The system then clusters the image dataset using our Pillar-Kmeans algorithm. The centroids of clustering results are used for calculating the similarity measurements to the image query. We perform our proposed system for experimental purpose with the Ukiyo-e image dataset from Tokyo Metropolitan Library for representing the Japanese cultural image collections.

AB - This paper presents an image search system with an emotion-oriented context recognition mechanism. Our motivation implementing an emotional context is to express user's impressions for retrieval process in the image search system. This emotional context recognizes the most important features by connecting the user's impressions to the image queries. The Mathematical Model of Meaning (MMM: [2], [4] and [5]) is applied for recognizing a series of emotional contexts for retrieving the most highly correlated impressions to the context. These impressions are then projected to a color impression metric to obtain the most significant colors for subspace feature selection. After applying subspace feature selection, the system then clusters the subspace color features of the image dataset using our proposed Pillar-Kmeans algorithm. Pillar algorithm is an algorithm to optimize the initial centroids for K-means clustering. This algorithm is very robust and superior for initial centroids optimization for K-means by positioning all centroids far separately among them in the data distribution. It is inspiring that by distributing the pillars as far as possible from each other within the pressure distribution of a roof, the pillars can withstand the roof's pressure and stabilize a house or building. It considers the pillars which should be located as far as possible from each other to withstand against the pressure distribution of a roof, as number of centroids among the gravity weight of data distribution in the vector space. Therefore, this algorithm designates positions of initial centroids in the farthest accumulated distance between them in the data distribution. The cluster based similarity measurement also involves a semantic filtering mechanism. This mechanism filters out the unimportant image data items to the context in order to speed up the computational execution for image search process. The system then clusters the image dataset using our Pillar-Kmeans algorithm. The centroids of clustering results are used for calculating the similarity measurements to the image query. We perform our proposed system for experimental purpose with the Ukiyo-e image dataset from Tokyo Metropolitan Library for representing the Japanese cultural image collections.

KW - cluster based similarity

KW - emotional context

KW - Image search

KW - multi-query images

KW - subspace feature selection

UR - http://www.scopus.com/inward/record.url?scp=79951601930&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79951601930&partnerID=8YFLogxK

U2 - 10.3233/978-1-60750-690-4-117

DO - 10.3233/978-1-60750-690-4-117

M3 - Conference contribution

SN - 9781607506898

VL - 225

T3 - Frontiers in Artificial Intelligence and Applications

SP - 117

EP - 136

BT - Frontiers in Artificial Intelligence and Applications

ER -