Dynamic License Plate Recognition Based on Dynamic Image Processing Technology
| Autor: | Zong-Lin Yang, 楊宗霖 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 There are two types of current license plate recognition methods, called static and dynamic. For static identification, the identification material is based on only a single caught image. Therefore, it may be affected by the environment, such as the lighting strength and angle, or the weather when the image caught. These factors may lead to recognition errors or unrecognizable. Dynamic identification, which uses of multiple sheets in a row of images to identify, may reduce the environmental impact factors. However, due to the different traffic conditions of each road, it may result in different recognition rate. If the license plate is too small or too much screen objects in a frame, the positioning accuracy may be significantly decreased. Furthermore, due to the identification of multiple consecutive of images at the same time, if the computer used is incapable of processing such huge amount of data, the identification cannot be completed immediately. If you reduce the number of images in a row retrieved, it will result in identification efficiency is reduced, thereby affecting the recognition results. In order to solve the identification problems above, this study proposes a recognition technique based on dynamic image processing techniques. In order to reduce the screen interference, users can follow the results of the road tests and to adjust the detection frame size and to improve the accuracy of license plate localization and thus enhance the identification efficiency and recognition results. Identification, in order to avoid the effects of the environment or picture identification result, the system will identify prospective heart after each frame license plate frame, use repetition rate calculation methods, the recognition results of each frame, remove duplicate the highest degree of recognition as a result of the license plate, the results of repeated use of multi-way, to address the environmental impacts of the screen or cause the failure does not recognize or identify the problem. In addition, the user can select direct inputs or video inputs. Using the direct inputs, the system can recognize the license plates in real-time through the camera or Webcam. The user can input the video file inputs to the system, to achieve the purpose of finding some specific car licenses. After users adjust the system based on some road condition parameters, an effective solution to the license plate recognition can enhance the overall recognition rate from 94.88% to 100%, except the defaced license plates. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|