Dissertation > Excellent graduate degree dissertation topics show

Research on Application of visual saliency computation in natural images and interested in target detection

Author: CaiWanLi
Tutor: ZhaoGaoPeng
School: Nanjing University of Technology and Engineering
Course: Control Theory and Control Engineering
Keywords: Visual saliency Interested object detection Global contrast Candidate region
CLC: TP391.41
Type: Master's thesis
Year: 2014
Downloads: 41
Quote: 0
Read: Download Dissertation

Abstract


Visual attention means that when human beings face the complex scenes, they will focus the important area in the scene quickly, and handle it in priority. Nowadays, Image processing takes a lot of time and computing power when dealing with the complex scenes. Therefore, it is very necessary to imitate the Human Visual System to establish a visual model, which has important applications in Interested Object Detection, Object Recognition, Image Retrieval, Image and Video Compression etc. So it is very necessary to research it.In the field of computer vision, the research of visual attention is becoming a hot spot. So we propose two visual computing methods on the basis of existing models, and use it in the interested object detection for natural images. The main contents of this paper are as follows:(1) The status of visual saliency is stated in detail, the advantages and disadvantages of existing models are analysed, and the basics of visual saliency is explained.(2) Propose a visual saliency method based global contrast in connection with the salient feature maybe different through the different color component. According to the color of salient regions is rare in the global picture, and it is centralized in the spatial distribution. We obtain the salient picture both in the HSV space and the Lab space. And then merge them to obtain the final salient image. Then we use a large number of images to analyse the method, the result indicates that this method is better than several typical methods.(3) Propose a visual saliency method based the candidate area in connection with the disadvantage of obtaining the salient area, which depend on the salient picture totally. We use the method of hierarchical image segmentation to generate candidate region, and computing the salient value through the spatial domain, frequency domain, closed contours and the local color contrast. Then we get the salient area through sorting the candidate area. We use a large number of pictures to analyse the method. And the result indicates that this method is better than several typical methods.(4) Through improve the visual saliency models, we use them in the field of object detection. Firstly we divide the natural image into six classes:Sky, River, Desert, Grass, Forest and City background. We make an experiment and analysis the method. The results show the effectiveness of our method.

Related Dissertations

  1. The Research on Key Technology of Computational Model for Visual Attention,TP391.41
  2. Saliency Inspired Objective Quality Model on Packet-Loss-Impired Image and Video,TP391.41
  3. Studies on Visual Pattern Mining Based on Web Images,TP391.41
  4. Research on Vision Based Mapping and Localization for Mobile Robot in Unknown Environments,TP242
  5. Top-Down Attention Motivated Research on Perception Model,P208
  6. Visual Attention-Based Airport Detection in Remote Sensing Images,TP751
  7. Research on Image Salient Region Detection Methods And Applications,TP391.41
  8. Research on Applications Based on Visual Saliency,TP391.41
  9. Study on the Optimization of Image Content and Structure,TP391.41
  10. Evaluation method of dual scale edge structural similarity image quality based on ROI,TP391.41
  11. Shape Skeleton Extraction and Shape Decomposition Based on Visual Saliency Features,TP391.41
  12. Saliency Detection Algorithm Based on Space Comparison and Analysis of Visual Feature,TP391.41
  13. Visual Attention Model and Its Application on Scene Categorization,TP391.41
  14. Research and Implementation of Pedestrian Detection Based on Monocular Vision in the Scene of Car Backing,TP391.41
  15. Research on detection and tobacco leaf stem type recognition method based on machine vision,TP391.41
  16. Object Detection Based on Visual Attention Models,TP391.41
  17. Salient Region Detection Based on Color Contrast and Anisotropic Diffusion Segmentation,TP391.41
  18. The Research on HITS-based Image Saliency Detection,TP391.41
  19. Traiffc Sign Detection and Recognition Research Based on Visual Saliency,TP391.41
  20. The Design and Implementation of Distributed Motion Detection and Tracking System,TP391.41
  21. Visual Saliency Based Objects Recognition,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