Dissertation > Excellent graduate degree dissertation topics show

Research on Algorithms to Reassign Labels to Regions

Author: ZuoZhou
Tutor: GuoYueFei
School: Fudan University
Course: Applied Computer Technology
Keywords: Image Processing Content-based Image Retrieval SIFT Bag of words model Maximum expected
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 117
Quote: 0
Read: Download Dissertation

Abstract


With today's rapid development of Internet, a variety of different forms, different content geometric image data is also orders of magnitude faster surge. Faced with such a diverse and vast image database, how fast and accurate image retrieval results to customer satisfaction has become a practical and urgent need to solve the problem. Efficient image retrieval techniques can greatly help people in the digital entertainment on the Internet, to improve people's quality of life. Current web-based text retrieval technology is relatively already quite mature, but Google, Baidu, and a variety of other companies Flickr image-based retrieval techniques but because of their own shortcomings are still far from meeting the needs of users. More concerned about the academic content-based image retrieval technology is mainly on account of the visual image itself by mining semantic characteristics related searches. From the image, we can extract the color, texture, shape, and a variety of other key visual features, and then we'll use an image similarity calculation or the use of pattern recognition and machine learning methods to analyze these images contain high semantic information, and finally integration of various other algorithms to retrieve relevant results. While there are many methods used to train the learning image retrieval engine, the relevant aspects of the study also had several decades of development, but the current content-based image retrieval technology to achieve the performance is not very satisfactory. On the one hand mainly perceived divide exists, use current technology to extract visual features for image content expression is not enough, and these are content-based image retrieval technology based on the other hand lies semantic gap, does not yet have A better approach may be standardized by different people with an understanding of image expression. Therefore, the current focus is on how best to extract and represent the image itself contains information on various characteristics, as well as how to properly use feature information to help us realize the link between content-based image retrieval. In this paper, we focus on the image based on multi-label learning algorithm for automatic annotation semantic region, with the ultimate aim is to improve the image retrieval accuracy and efficiency. We propose an iterative EM-based multi-label area unsupervised image calibration algorithm, it can be very effective in the label based on full automatic calibration graph corresponds to the image of the local area. We first of all images densely sampled SIFT feature points, and then get mature in the field of text processing applications bag of words used to model analog content-based image classification, the combination of all the SIFT feature points obtained K-means clustering image visual dictionaries, and then construct an iterative process greatest expectations algorithm calculates each image in the presence of each label for each image visual WORD confidence, and finally select those images with a higher degree of confidence visual WORD, identify each image for each tag area corresponding to the highest confidence. Experiments show that, in the case of the full sample data, we propose an algorithm to solve unsupervised automatic calibration, labeling the apparent diversity and multi-label and other issues have achieved good results. Follow-up work will be carried out primarily focused on improving feature representation of diversity and effective feature combinations, thereby improving the algorithm more kinds of labels on the applicability and accuracy. Finally, we are the future of content-based image retrieval technology to make the prospect.

Related Dissertations

  1. Research on Temperature Measurement Technique Based on CCD Image Sensor,TH811
  2. Research on Basic Algorithms of Digital Image Processing and Implementation with FPGA,TP391.41
  3. The Research and Implemention of Image Retrieval Based on User Interested Feature,TP391.41
  4. Research of Images Enhancing Algorithms on Fog or Backlighting Conditions and Implementation with Hardwares,TP391.41
  5. Application of Q-Learning in the Content-Based Image Retrieval Technology,TP391.41
  6. Research and Implementation on Content-Based Clothing Image Retrieval,TP391.41
  7. The Discovery of User Concept Region Based on Multiple Instance Learning,TP391.41
  8. Research on the Image Real-Time Acquisition, Storage and Image Processing System,TP391.41
  9. The Research of Moving Object Tracking System Based on Embeded Image Process Unit,TP391.41
  10. Research on Visual Servo System of Mechanical ARM,TP242.6
  11. Application Research of Digital Image Processing on Container Inspection,TP274.4
  12. Three-Dimensional Numerical Modelling for the Head and the Electric Field Analysis of DBS,R742.5
  13. Application of Semi-Structure in Leisure Dress,TS941.2
  14. Research on Nondestructive Detection Technology for External Qualities of Papayas Based-on Vision,S667.9
  15. Research on Detecting Optical Fiber Geometric Parameter Based on Machine Vision,TN253
  16. Research on Identification System of Cashmere and Wool Fiber,TS101.921
  17. Design of Garlic Planting Sorting Machine Based on Image Processing,S223.2
  18. Study on Detection and Grading of ’Jiro’ Persimmon’s External Quality Based on Computer Vision,S665.2
  19. Flow Behavior Description of Two-Phase Flow Based on Image Processing Techniques,TP391.41
  20. Active faults based radar image information extraction method applied research and demonstration,P542.3
  21. Research on License Plate Location Algorithm of License Plate Recognition System,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile