国产bbaaaaa片,成年美女黄网站色视频免费,成年黄大片,а天堂中文最新一区二区三区,成人精品视频一区二区三区尤物

首頁> 外文學(xué)位 >Automated license plate recognition: A novel approach using spectral analysis and majority vote neural networks.
【24h】

Automated license plate recognition: A novel approach using spectral analysis and majority vote neural networks.

機(jī)譯：自動車牌識別：一種使用頻譜分析和多數(shù)表決神經(jīng)網(wǎng)絡(luò)的新穎方法。

獲取原文
獲取原文并翻譯 | 示例
  • 摘要
  • 著錄項(xiàng)
  • 相似文獻(xiàn)
  • 相關(guān)主題

摘要

This thesis presents an automated License Plate Recognition (LPR) system for United States license plates as an integrated module to assist vehicle emission measurement study. The recognition problem is divided into three stages: extraction of the license plate from the vehicle image, segmentation of characters on the license plate and recognition of the segmented characters. The extraction of the license plate image and the subsequent stage of character segmentation are based on power spectral density (PSD) analysis using periodogram estimation techniques. The Optical Character Recognition (OCR) system implemented for recognizing the segmented characters incorporates a novel feature extraction methodology based on character edges and employs a majority vote neural network for classification. The features used for recognition are the edge pattern of characters generated by scanning the binary image of the character horizontally, vertically, and diagonally. The algorithms implemented at each stage have been tested on a database of few hundred license plate images captured in vehicle emission measurement studies and is found to exhibit a success rate of more than 90%. It is shown that the developed methodologies are very robust to degradation in license plate images owing to factors such as dust, ambient illumination, speed of the vehicle and poor image resolution.
機(jī)譯：本文提出了一種用于美國車牌的自動車牌識別（LPR）系統(tǒng)，作為一個(gè)集成模塊來輔助車輛排放測量研究。識別問題分為三個(gè)階段：從車輛圖像中提取車牌，對車牌上的字符進(jìn)行分割以及對分割后的字符進(jìn)行識別。車牌圖像的提取和字符分割的后續(xù)階段基于使用周期圖估計(jì)技術(shù)的功率譜密度（PSD）分析。用于識別分段字符的光學(xué)字符識別（OCR）系統(tǒng)結(jié)合了基于字符邊緣的新穎特征提取方法，并采用多數(shù)表決神經(jīng)網(wǎng)絡(luò)進(jìn)行分類。用于識別的特征是通過水平，垂直和對角掃描字符的二進(jìn)制圖像而生成的字符的邊緣圖案。已在車輛排放測量研究中捕獲的數(shù)百個(gè)車牌圖像的數(shù)據(jù)庫中對在每個(gè)階段實(shí)施的算法進(jìn)行了測試，發(fā)現(xiàn)其成功率超過90％。結(jié)果表明，由于諸如灰塵，環(huán)境照明，車輛速度和較差的圖像分辨率等因素，所開發(fā)的方法對于車牌圖像的退化非常有力。

著錄項(xiàng)

相似文獻(xiàn)

  • 外文文獻(xiàn)
  • 中文文獻(xiàn)
  • 專利
獲取原文

客服郵箱：kefu@zhangqiaokeyan.com

京公網(wǎng)安備：11010802029741號 ICP備案號：京ICP備15016152號-6 六維聯(lián)合信息科技 (北京) 有限公司?版權(quán)所有

  • 客服微信

  • 服務(wù)號