Presentation is loading. Please wait.

Presentation is loading. Please wait.

指導老師 : 蔡亮宙 報告者 : 黃柏愷 A new method of vehicle license plate location under complex scenes.

Similar presentations


Presentation on theme: "指導老師 : 蔡亮宙 報告者 : 黃柏愷 A new method of vehicle license plate location under complex scenes."— Presentation transcript:

1 指導老師 : 蔡亮宙 報告者 : 黃柏愷 A new method of vehicle license plate location under complex scenes

2 Outline INTRODUCTION LICENSE PLATE LOCATION SYSTEM Image pre-processing Image horizontal differences Gradient Image adaptive threshold segmentation Morphological operations Region merging The achievement of positioning through regional search EXPERIMENT AND ANALYSIS CONCLUSIONS REFERENCES

3 INTRODUCTION A LPR(license plate recognition) system is mainly composed of three processing modules: license plate location character segmentation character recognition Locate various kinds of license plates in a short time and meet the needs of real-time location of vehicles

4 LICENSE PLATE LOCATION SYSTEM The flowchart of license plate location

5 Image pre-processing Gray-scale processing of the images

6 Image horizontal differences gray gradient image G Gray gradient image

7 Gradient Image adaptive threshold segmentation(1/4) Steps are as follows the initial weights factor γ =0.6 Weight increment factor δ =0.05 Gw is the maximum of every row in gradient image γ as experience value and needs no modification (a)

8 Gradient Image adaptive threshold segmentation(2/4) (b) Segment G by making use of T

9 Gradient Image adaptive threshold segmentation(3/4) (c) 1.opening operation of G’ by using 10x10 structuring element and eliminate some noise points 2.connected region of license plate can be formed through closure operation by utilizing 15×15 structuring element (d) Calculate the density of white points in the image w=Count/(H*W)

10 Gradient Image adaptive threshold segmentation(4/4) (e) Adopt ó1=0.03 ó2=0.08.If w<ó1,then γ = γ + δ ;if w>ó2,then γ = γ - δ. The algorithm turn to step a),otherwise, the process ends

11 Morphological operations further morphological operation to reduce the number of irrelevant regions Steps are as follows Small noise regions are dispelled through morphological opening operation with 20x20 structural elements erosion and inflation operation are conducted by using 6x6 structural elements Dilation is conducted by making use of 25x5 structural elements

12 Region merging The principle of region merging is:if the absolute offset of rows and columns in two neighbor regions are within certain range (a)The result of further morphological processing (b)The result of region merging (a)(b)

13 The achievement of positioning through regional search Steps are as follows the aspect ratio is between 2.2 and 3.5 background color and word color, we make binaryzation of every candidate region with the method of global Otsu method proportion of target points in the overall candidate region ; this proportion is usually between 0.25 and 0.65 w=count/(h*w)

14 EXPERIMENT AND ANALYSIS(1/2) REGIONS EXPERIMENT RESULTS

15 EXPERIMENT AND ANALYSIS(2/2) The results of license plate location under different conditions

16 CONCLUSIONS advantage on the location of low-quality license plate shot in the complex conditions of changing light, noisy environment and far distances

17 REFERENCES [1] Wenju Li, Dequn Liang and Qi Zhang, “A novel approach for vehicle license plate location based on edge-color pair,” Chinese Journal of Computer, vol.27,2004,pp.204-208. [2] Jie Guo and Pengfei Shi, “Color and texture analysis based vehicle license plate location,” Journal of Image and Graphics,vol.7,2002,pp.426-427. [3] Park S H,Kim K I and Jung K, “Locating car license plates using neural networks,”Electronics Letters,vol.35,1999,pp.1475-1477. [4] Weiqi Yuan and Liang Zhang,“An improved vehicle plate location method based on block segmentation,”, ACTA AUTOMATICA SINICA,vol.33, 2007,pp.768-770. [5] Ichen Tsai, Juichen Wu, “Recognition of Vehicle License Plates from a Video Sequence,” IAENG International Journal of Computer Science,vol.36,2009. [6] Hsieh J W,Yu S H and Chen Y S, “Morphology based license plate detection from complex scenes,”Proceedings of 16th International Conference on Pattern Recognition(ICPR‘02),IEEE Press,2002,pp.176-179. [7] Yue Lu,Dongjian He and Xiao He, “Fast algorithm for vehicle license plate localization based on scan line and feature selection,”Computer Engineering and Design,vol.29,2008,pp.5125-5128.


Download ppt "指導老師 : 蔡亮宙 報告者 : 黃柏愷 A new method of vehicle license plate location under complex scenes."

Similar presentations


Ads by Google