Novelty detection is an important functionality that has found many applications in information r... more Novelty detection is an important functionality that has found many applications in information retrieval and processing. In this paper we propose a novel framework that deals with novelty detection in multiple-scene image sets. Working with wildlife image data, the framework starts with image segmentation, followed by feature extraction and classification of the image blocks extracted from image segments. The labelled image blocks are then scanned through to generate a co-occurrence matrix of object labels, representing the semantic context within the scene. The semantic co-occurrence matrices then undergo binarization and principal component analysis for dimension reduction, forming the basis for constructing one-class models on scene categories. An algorithm for outliers detection that employs multiple one-class models is proposed. An advantage of our approach is that it can be used for novelty detection and scene classification at the same time. Our experiments show that the proposed approach algorithm gives favourable performance for the task of detecting novel wildlife scenes, and binarization of the semantic co-occurrence matrices helps increase the robustness to variations of scene statistics.► We propose to construct co-occurrence matrices on semantics level as a new feature. ► Statistical tests reveal good capabilities of the new feature for novelty modelling. ► Mahalanobis distance applied on the feature produces favourable performance. ► A multiple one-class classification model is used for novelty and scene classification.
There is a growing evidence that visual saliency can be better modeled using top-down mechanisms ... more There is a growing evidence that visual saliency can be better modeled using top-down mechanisms that incorporate object semantics. This suggests a new direction for image and video analysis, where semantics extraction can be effectively utilized to improve video summarization, indexing and retrieval. This paper presents a framework that models semantic contexts for key-frame extraction. Semantic context of video frames is extracted and its sequential changes are monitored so that significant novelties are located using a one-class classifier. Working with wildlife video frames, the framework undergoes image segmentation, feature extraction and matching of image blocks, and then a co-occurrence matrix of semantic labels is constructed to represent the semantic context within the scene. Experiments show that our approach using high-level semantic modeling achieves better key-frame extraction as compared with its counterparts using low-level features.
Page 21. Localized Adaptive Learning of Mixture of Gaussians Models for Background Extraction (LA... more Page 21. Localized Adaptive Learning of Mixture of Gaussians Models for Background Extraction (LAL-MoG) Munir Shah Department of information Science, University of Otago mshah@ infoscience. otago. ac. nz Jeremiah Deng ...
Innovations in Intelligent Image Analysis, Jan 1, 2011
Abstract. Feature extraction and selection has always been an interesting is-sue for pattern reco... more Abstract. Feature extraction and selection has always been an interesting is-sue for pattern recognition tasks. There have been numerous feature schemes proposed and empirically validated for image scene and object categorization problems, no matter it is for general-purposed ...
Image and Vision Computing New Zealand, 2009. …, Jan 1, 2009
... image categorization Jeremiah D. Deng Department of Information Science University of Otago, ... more ... image categorization Jeremiah D. Deng Department of Information Science University of Otago, PO Box 56, Dunedin 9054, New Zealand E-mail: jeremiah.deng@otago.ac.nz AbstractMedical image annotation remains a challenging task. ...
Journal of Network and Computer Applications, Jan 1, 2011
Recently, performance optimization on multi-core platforms has received more and more research in... more Recently, performance optimization on multi-core platforms has received more and more research interest especially for networking applications. While a number of case studies have been made with success, there is however a lack of analytical treatment on the modeling of parallel processing in applications. In this paper, we tackle the performance optimization problem using queueing analysis. Modeling macroscopic pipeline processes with a constrained tandem model, an optimization principle is derived. We then employ the derived principle to analyze two network applications: intrusion detection and image retrieval. Both theoretical performance analysis and simulation results confirm the effectiveness of our optimization principle.
What statistics makes an image of natural scene different from that of an urban or artificial env... more What statistics makes an image of natural scene different from that of an urban or artificial environment? This has been an intriguing research topic in recent years. In this paper, in contrast to the more popular approach of using global frequency spectral features, we propose to employ a few statistics measures derived from local edge and fractal features. Feature selection methods are used to reveal the relevance of the defined features. The effectiveness of the proposed feature scheme is demonstrated using benchmark digital image data as well as high dynamic range image data.
Proceedings of the 20th Australian joint …, Jan 1, 2007
Sapstain is caused by fungi that can infect and subsequently damage the wood quality. Timber can ... more Sapstain is caused by fungi that can infect and subsequently damage the wood quality. Timber can be infected at any time during manufacturer if the timber becomes moist enough. Sapstain can penetrate deep into the timber, which often means the removing the surface stain will ...
Novelty detection is an important functionality that has found many applications in information r... more Novelty detection is an important functionality that has found many applications in information retrieval and processing. In this paper we propose a novel framework that deals with novelty detection in multiple-scene image sets. Working with wildlife image data, the framework starts with image segmentation, followed by feature extraction and classification of the image blocks extracted from image segments. The labelled image blocks are then scanned through to generate a co-occurrence matrix of object labels, representing the semantic context within the scene. The semantic co-occurrence matrices then undergo binarization and principal component analysis for dimension reduction, forming the basis for constructing one-class models on scene categories. An algorithm for outliers detection that employs multiple one-class models is proposed. An advantage of our approach is that it can be used for novelty detection and scene classification at the same time. Our experiments show that the proposed approach algorithm gives favourable performance for the task of detecting novel wildlife scenes, and binarization of the semantic co-occurrence matrices helps increase the robustness to variations of scene statistics.► We propose to construct co-occurrence matrices on semantics level as a new feature. ► Statistical tests reveal good capabilities of the new feature for novelty modelling. ► Mahalanobis distance applied on the feature produces favourable performance. ► A multiple one-class classification model is used for novelty and scene classification.
There is a growing evidence that visual saliency can be better modeled using top-down mechanisms ... more There is a growing evidence that visual saliency can be better modeled using top-down mechanisms that incorporate object semantics. This suggests a new direction for image and video analysis, where semantics extraction can be effectively utilized to improve video summarization, indexing and retrieval. This paper presents a framework that models semantic contexts for key-frame extraction. Semantic context of video frames is extracted and its sequential changes are monitored so that significant novelties are located using a one-class classifier. Working with wildlife video frames, the framework undergoes image segmentation, feature extraction and matching of image blocks, and then a co-occurrence matrix of semantic labels is constructed to represent the semantic context within the scene. Experiments show that our approach using high-level semantic modeling achieves better key-frame extraction as compared with its counterparts using low-level features.
Page 21. Localized Adaptive Learning of Mixture of Gaussians Models for Background Extraction (LA... more Page 21. Localized Adaptive Learning of Mixture of Gaussians Models for Background Extraction (LAL-MoG) Munir Shah Department of information Science, University of Otago mshah@ infoscience. otago. ac. nz Jeremiah Deng ...
Innovations in Intelligent Image Analysis, Jan 1, 2011
Abstract. Feature extraction and selection has always been an interesting is-sue for pattern reco... more Abstract. Feature extraction and selection has always been an interesting is-sue for pattern recognition tasks. There have been numerous feature schemes proposed and empirically validated for image scene and object categorization problems, no matter it is for general-purposed ...
Image and Vision Computing New Zealand, 2009. …, Jan 1, 2009
... image categorization Jeremiah D. Deng Department of Information Science University of Otago, ... more ... image categorization Jeremiah D. Deng Department of Information Science University of Otago, PO Box 56, Dunedin 9054, New Zealand E-mail: jeremiah.deng@otago.ac.nz AbstractMedical image annotation remains a challenging task. ...
Journal of Network and Computer Applications, Jan 1, 2011
Recently, performance optimization on multi-core platforms has received more and more research in... more Recently, performance optimization on multi-core platforms has received more and more research interest especially for networking applications. While a number of case studies have been made with success, there is however a lack of analytical treatment on the modeling of parallel processing in applications. In this paper, we tackle the performance optimization problem using queueing analysis. Modeling macroscopic pipeline processes with a constrained tandem model, an optimization principle is derived. We then employ the derived principle to analyze two network applications: intrusion detection and image retrieval. Both theoretical performance analysis and simulation results confirm the effectiveness of our optimization principle.
What statistics makes an image of natural scene different from that of an urban or artificial env... more What statistics makes an image of natural scene different from that of an urban or artificial environment? This has been an intriguing research topic in recent years. In this paper, in contrast to the more popular approach of using global frequency spectral features, we propose to employ a few statistics measures derived from local edge and fractal features. Feature selection methods are used to reveal the relevance of the defined features. The effectiveness of the proposed feature scheme is demonstrated using benchmark digital image data as well as high dynamic range image data.
Proceedings of the 20th Australian joint …, Jan 1, 2007
Sapstain is caused by fungi that can infect and subsequently damage the wood quality. Timber can ... more Sapstain is caused by fungi that can infect and subsequently damage the wood quality. Timber can be infected at any time during manufacturer if the timber becomes moist enough. Sapstain can penetrate deep into the timber, which often means the removing the surface stain will ...
Uploads