Dissertation > Excellent graduate degree dissertation topics show

A Study of Car Detection in Highway with High Resolution Aerial Photo

Author: ZhengZeZhong
Tutor: FanDongMingï¼›ZhouGuoQing
School: Southwest Jiaotong University
Course: Cartography and Geographic Information Engineering
Keywords: Highway car target detection Varimax Edge Detection Template matching Grayscale mathematical morphology Binary mathematical morphology High - resolution aerial images
CLC: TP391.41
Type: PhD thesis
Year: 2010
Downloads: 174
Quote: 0
Read: Download Dissertation

Abstract


The auto target detection information collection aspects in the field of intelligent transportation research is a very important issue, based on a hot car aerial imagery target detection image processing. Unusually severe because of the rapid development of the city, the sharp increase of the number of cars, causing traffic jams. Therefore, in transportation planning, control and management of the program development process, and how to ensure the coordination of road transport network and urban development, and how to ensure the comprehensiveness of the traffic survey data and now potential resistance, optimize network structure, rational distribution of the urban transport network, the transport network can play a role fully and efficiently, is extremely important. Car target detection can just take full advantage of high-resolution aerial remote sensing image rich spatial information necessary car traffic information, traffic management department, in order to achieve the real rationalization of \In this thesis, using digital mosaic of high-resolution aerial imagery data, cutting on the basis of a typical highway image maximum variance method, edge detection, template matching, and grayscale mathematical morphology and binary mathematical morphology algorithm combined the car target detection, the paper around the car a target detection research work mainly as follows: (1) study the maximum variance (Otsu) method, threshold-based segmentation algorithm to automatically determine the optimal threshold, the high-resolution aerial Expressway image binarization, combined with binary mathematical morphological opening operation operating a vehicle target detection. Experimental results show that high detection rate of the algorithm for a simple background image; complex background images car target detection, low accuracy rate. (2) study of several typical edge detection binarization algorithm, combined with binary mathematical morphology algorithm automotive target detection. The experimental results show that, based on the Robert operator edge detection the binarization image edge continuity better based on Sobel operator, Prewitt operator the sub edge detection binarization image; based on Sobel operator, Prewitt operator Edge Detection binarization image effects not as good as the Laplace edge detection results binarization image and Canny edge detection results binarization image; Canny operator is detected in all edge detection operators the best. But the the car target detection results show that, simple background, Sobel operator promoter and binary mathematical morphology combined, the highest rate of auto target detection; complex background using Canny operator or Sobel operator and binary mathematical morphology combined automotive target detection works best, but complex background car target detection success rate is very low. (3) target detection based on template matching algorithm for high-resolution aerial images highway vehicles. Experimental results show that the high resolution aerial imagery, the details of the car target is clear, therefore, the template matching algorithm to detect vehicle targets lies in the establishment of various car brands, Vehicle Template Library. Template matching with varimax edge detection method compared to a huge amount of calculation; Meanwhile, many car brands, each brand car models also a lot to establish the workload of the template library is also very huge. (4) to study the combination of high-resolution aerial images highway vehicle target detection based on gray-scale mathematical morphology and binary mathematical morphology algorithm. Complex background, a combination of top-hat transform and open computing, to the bright background of the car, can detect target screening earth objects (dark background) and small surface features; combination of low cap transformation and closing operation, and screening small surface features, can be detected by the target of the car on a dark background; open computing and closing operation detected automotive target superimposed, and \The algorithm auto target detection harmonic mean (Fm) of 94% or more, and can achieve a good car target detection effect. (5) In general, the gray-scale mathematical morphology and binary mathematical morphology algorithm combined for automotive target detection, compared with the maximum variance method based edge detection algorithm, automotive target detection accuracy with more robustness, but the program running time is slightly longer. Grayscale mathematical morphology and binary mathematical morphology algorithm with template matching algorithm has strong robustness and efficiency.

Related Dissertations

  1. Research on Detecting Optical Fiber Geometric Parameter Based on Machine Vision,TN253
  2. Research on Abnormal Vehicle Behavior Detection and Application in Traffic Video,TP391.41
  3. Tracking Events for Food Complaint Documents Based on Ontology,TP391.1
  4. Research on Vision Detection Algorithm for Tracing Printing System,TP391.41
  5. Research on Image Segmentation Method Based on Measure of Medium Truth Degree,TP391.41
  6. Research on Noise Reducation and Color Interpolation Algorithms Based on Bayer Pattern Images for Digital Camera,TP391.41
  7. Research on Image Super-resolution Reconstruction Based on Non-local Similarity,TP391.41
  8. Visual Servoing Approaches Based on Parallel Mechanism,TP391.41
  9. Research of Image Segmentation in Web Image Search Based on GPUs,TP391.41
  10. Spectrum Information hyperspectral imaging target detection technique,TP391.41
  11. Visual underwater target detection and recognition,TP391.41
  12. High-speed detection and bayonet vehicle license identification information,TP391.41
  13. Image intensifier and a screen resolution of defect detection technology,TN144
  14. Wheel based on image processing envelope contour measurement principle and technology research,TP391.41
  15. Color medical image segmentation technology research,TP391.41
  16. Servers based on infrared image classification conditions,TP393.05
  17. Based on structural similarity with the MTF image quality evaluation method,TP391.41
  18. PDE-based image denoising and enhancement of,TP391.41
  19. Pattern recognition method based weather radar monitoring and analysis of digital products,TN959.4
  20. Based on Image Processing Method of automatic target,TP391.41
  21. Mobile robot vision -based target search and tracking technology,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