检索结果-内蒙古大学图书馆

A novel multiseed nonhierarchical data clustering technique

IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS 1997年第5期27卷 871-877页

作者： Chaudhuri, D Chaudhuri, BB Computer Vision and Pattern Recognition Unit Indian Statistical Institute Calcutta India

Clustering techniques such as K-means and Forgy as well as their improved version ISODATA group data around one seed point for each cluster. It is well known that these methods do not work well if the shape of the cluster is elongated or nonconvex. We argue that for a elongated or nonconvex shaped cluster, more than one seed is needed, In this paper a multiseed clustering algorithm is proposed. A density based representative point selection algorithm is used to choose the initial seed points. To assign several seed points to one cluster, a minimal spanning tree guided novel technique is proposed. Also, a border point detection algorithm is proposed for the detection of shape of the cluster. This border in turn signifies whether the cluster is elongated or not. Experimental results show the efficiency of this clustering technique.

关键词： classification minimal spanning tree multiseed clustering pattern recognition

来源：评论

学校读者我要写书评

暂无评论

Multi-scale integration with semantic embedding and adaptive excitation transformer for underwater optical image enhancement

引用

Optics and Laser Technology 2025年 189卷

作者： Yang, Jing Liang, Hui Zhu, Shanbing Bai, Shumin Jiang, Fengling Hussain, Amir School of Artificial Intelligence and Big Data Hefei University Hefei230601 China Collaborative Innovation Laboratory for Computer Vision and Pattern Recognition Hefei University Hefei230601 China Hefei Normal University Hefei230601 China School of Computing Edinburgh Napier University Merchiston Campus Scotland EdinburghEH10 5DT United Kingdom

As an important technology in the domains of ocean exploration and underwater robotics, underwater optical image enhancement has drawn significant attention. However, underwater images suffer from severe degradation due to wavelength-related optical attenuation, refraction, and scattering. The need to acquire high-quality underwater optical images poses challenges to current techniques. A dual-branch neural network based on multi-scale integration with semantic embedding and adaptive excitation transformer for underwater optical image enhancement is proposed in this paper. First, a semantic embedding network serving as the semantic branch is introduced to extract low- and high-level semantic features effectively to generate images with clearer edges and texture details. Concurrently, an enhancement branch containing adaptive excitation transformer is constructed to enhance local details while concentrating on global information. Since different feature channels correspond to different patterns of the input image, an adaptive excitation mechanism is deployed to achieve adaptive estimations of channel weights, highlighting the more representative channels while penalizing less important channels to enrich texture and eliminate blurring and color deviations. Additionally, a multi-scale dynamic integration module is designed. It establishes a correlation between the semantic and enhancement branches and adaptively selects prominent features to avoid over-enhancement and improve clarity and color deviation. Both the qualitative and quantitative results evaluated on public underwater optical image datasets show that our method outperforms the state-of-the-art methods in terms of subjective perception and evaluation metrics, indicating outstanding learning and generalization capabilities. Furthermore, excellent performance highlights the substantial benefits it contributes to downstream visual-related engineering tasks. © 2025 Elsevier Ltd

关键词： Adaptive optics

来源：评论

学校读者我要写书评

暂无评论

Texture synthesis by a neural network model

引用

NEURAL COMPUTING & APPLICATIONS 1997年第1期6卷 2-11页

作者： Chaudhuri, BB Kundu, P Computer Vision and Pattern Recognition Unit Indian Statistical Institute Calcutta India

In this paper we propose a neural network model to synthesise texture images. The model is based on a continuous Hopfield-like network where each pixel of the image is occupied by a neuron that is eight-connected to its neighbours. A state of the neuron denotes a certain grey level of the corresponding pixel. The firing of the neuron changes its state, and hence the grey level of the corresponding pixel. Different two-tone and grey-tone texture images can be synthesised by manipulating the connection weights and by varying the algorithm iteration number. For grey-tone texture synthesis, a Markov chain principle has been employed to decide on the multiple state transition of a neuron. The model can be employed for texture propagation with the advantage that it allows propagation without showing any blocky effect.

关键词： computer graphics image processing texture synthesis

来源：评论

学校读者我要写书评

暂无评论

Morphological processing of Indian languages for lexical interaction with application to spelling error correction

引用

SADHANA-ACADEMY PROCEEDINGS IN ENGINEERING SCIENCES 1996年第3期21卷 363-380页

作者： Sengupta, P Chaudhuri, BB Computer Vision and Pattern Recognition Unit Indian Statistical Institute Calcutta India

An NLP system for Indian languages should have a lexical subsystem that is driven by a morphological analyzer. Such an analyzer should be able to parse a word into its constituent morphemes and obtain lexical projection of the word as a unification of the projections of the constituent morphemes. Lexical projections considered here are f-structures of the Lexical Functional Grammar (LFG). A formalism has been proposed, by which the lexicon writer may specify the lexicon in four levels. The specifications are compiled into a stored lexical knowledge base on one hand and a formulation of derivational morphology called Augmented Finite State Automata (AFSA) on the other to achieve a compact lexical representation. The aspects of AFSA, especially its power of morphological parsing of words in a computationally attractive manner, has been discussed. An additional utility of the AFSA, in the form of spelling error corrector, has also been discussed. Bangla, or Bengali is considered as a case study. Implementation notes based on object-oriented programming principles has been provided.

关键词： natural language processing morphological sub-system lexical representation augmented finite state automata spelling corrector object-oriented implementation

来源：评论

学校读者我要写书评

暂无评论

LVAgent: Long Video Understanding by Multi-Round Dynamical Collaboration of MLLM Agents

arXiv

引用

arXiv 2025年

作者： Chen, Boyu Yue, Zhengrong Chen, Siran Wang, Zikang Liu, Yang Li, Peng Wang, Yali Shenzhen Key Lab of Computer Vision and Pattern Recognition Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences China School of Artificial Intelligence University of Chinese Academy of Sciences China Tsinghua University Beijing China Dept. of Comp. Sci. & Tech. Institute for AI Tsinghua University Beijing China Shanghai Artificial Intelligence Laboratory China Shanghai Jiao Tong University China

Existing Multimodal Large Language Models (MLLMs) encounter significant challenges in modeling the temporal context within long videos. Currently, mainstream Agent-based methods use external tools (e.g., search engine, memory banks, OCR, retrieval models) to assist a single MLLM in answering long video questions. Despite such tool-based support, a solitary MLLM still offers only a partial understanding of long videos, resulting in limited performance. In order to better address long video tasks, we introduce LVAgent, the first framework enabling multi-round dynamic collaboration of MLLM agents in long video understanding. Our methodology consists of four key steps: 1) Selection: We pre-select appropriate agents from the model library to form optimal agent teams based on different tasks. 2) Perception: We design an effective retrieval scheme for long videos, improving the coverage of critical temporal segments while maintaining computational efficiency. 3) Action: Agents answer long video-related questions and exchange reasons. 4) Reflection: We evaluate each agent’s performance in each round of discussion and optimize the agent team for dynamic collaboration. The agents iteratively refine their answers by multi-round dynamical collaboration of MLLM agents. LVAgent is the first agent system method that outperforms all closed-source models (including GPT-4o) and open-source models (including InternVL-2.5 and Qwen2-VL) in the long video understanding tasks. Our LVAgent achieves an accuracy of 80% on four mainstream long video understanding tasks. Notably, on the LongVideoBench dataset, LVAgent improves accuracy by up to 14.3% compared with SOTA. © 2025, CC BY-NC-SA.

关键词： Open systems

来源：评论

学校读者我要写书评

暂无评论

"Connected to the people": Social Inclusion & Cohesion in Action through a Cultural Heritage Digital Tool

引用

Proceedings of the ACM on Human-computer Interaction 2023年第CSCW2期7卷 1-37页

作者： Nisi, Valentina Bala, Paulo Cesário, Vanessa James, Stuart Del Bue, Alessio Nunes, Nuno Jardim Instituto Superior Técnico U. Lisbon Lisbon Portugal Visual Geometry and Modelling Lab Pattern Analysis and Computer Vision Istituto Italiano di Tecnologia Genova Italy Pattern Analysis and Computer Vision Istituto Italiano di Tecnologia Genova Italy

Current cultural policies are evolving from social inclusion (removing barriers and promoting equality for participation in culture) to social cohesion (fostering solid bonds between groups despite their differences). Digital interventions can create spaces that promote social inclusion and cohesion. In this paper, we report on the design and evaluation of a cultural heritage and digital storytelling application supporting a participatory approach to culture and hosting society. We evaluate our intervention in three marginalized communities with different social-cultural contexts: migrant women in Barcelona, a community living in a priority neighbourhood in Paris and second and third-generation migrants in Lisbon. Through an analysis of their application use, our findings point at their needs and desires, highlighting how the app can support social inclusion as the first step towards cohesion, but that these are heterogeneous concepts susceptible to nuanced appropriations by the different communities. © 2023 ACM.

关键词： Digital devices

来源：评论

学校读者我要写书评

暂无评论

An improved backpropagation neural network for detection of road-like features in satellite imagery

引用

INTERNATIONAL JOURNAL OF REMOTE SENSING 1997年第16期18卷 3379-3394页

作者： Bhattacharya, U Parui, SK Computer Vision and Pattern Recognition Unit Indian Statistical Institute Calcutta 700035 203 B. T. Road India

This paper presents an application of backpropagation neural network for the detection of linear structures in remote-sensing images. The purpose of the approach is two-fold. First, to exploit the advantages of a neural network classifier over the tranditional ones. Second, to avoid the strategic phases of enhancement and thresholding. Once the network is learnt, the classification scheme is real-time. Two critical issues in the present approach an the selection of the network architecture and the rate of convergence of learning. Solutions to these two problems are proposed. Experimental results on IRS and SPOT images are presented. Satisfactory classification results have been obtained using the network.

关键词： Remote sensing

来源：评论

学校读者我要写书评

暂无评论

A new definition of neighborhood of a point in multi-dimensional space

引用

pattern RECOGNITION LETTERS 1996年第1期17卷 11-17页

作者： Chaudhuri, BB Computer Vision and Pattern Recognition Unit Indian Statistical Institute 203 B.T. Road Calcutta 700 035 India

Given a set of points in multi-dimensional space, we propose a new definition for the neighbors of an arbitrary point P. The definition tries to capture the idea that the neighbors should be as near to P and as symmetrically placed around P as possible. In contrast, the conventional nearest neighborhood considers only nearness as the criterion for neighborhood. We propose an iterative procedure to compute the neighbors where the first neighbor is the nearest neighbor. The second and other neighbors are chosen so that at any stage the distance between the centroid of the neighbors and P is as small as possible. The centroid criterion takes care of symmetrical placement of the neighbors. One can use median instead of centroid to define the neighbors. The new definition is free from any user-specified parameter and can be used for pattern classification, clustering and low-level description of dot patterns.

关键词： neighborhood classification clustering pattern recognition image processing

来源：评论

学校读者我要写书评

暂无评论

PROPERTIES OF FRACTAL INTENSITY SURFACES

引用

pattern RECOGNITION LETTERS 1988年第5期8卷 319-324页

作者： GARDING, J Computer Vision and Association Pattern Processing Laboratory (CVAP) Royal Institute of Technology S-100 44 Stockholm Sweden

In this article we present an extended two-parameter version of the fractal Brownian model and investigate its application to computer vision tasks such as texture discrimination. Furthermore, we demonstrate that frac... 详细信息

关键词： Fractional Brownian functions fractal texture image segmentation

来源：评论

学校读者我要写书评

暂无评论

computer RECOGNITION OF PRINTED BANGLA SCRIPT

引用

INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE 1995年第11期26卷 2107-2123页

作者： PAL, U CHAUDHURI, BB Computer Vision and Pattern Recognition Unit Indian Statistical Institute Calcutta 700 035 203 B. T. Road India

This paper considers optical character recognition (OCR) of Bangla, the second most popular script in the Indian subcontinent. A complete OCR system is described for documents of single Bangla font, where more than three hundred character shapes are recognized by a combination of template and feature-matching approach. Here the document image captured by a flatbed scanner is subject to tilt correction, line, word and character segmentation, simple and compound character separation, feature extraction and finally character recognition. Some character occurrence statistics have been computed to aid the recognition process. The simple character recognition is done by a feature-based tree classifier, and the compound character recognition involves a template matching approach preceded by a feature-based grouping. At present, recognition accuracy of about 96% is obtained by the system.

关键词： Optical character recognition

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：