基于数字图像处理的车牌号码识别外文文献

基于数字图像处理的车牌号码识别外文文献内容摘要：

在计算机科学讲义：模式识别与支持向量机，页 293309， 2020 年国际研讨会。 [14]四洛韦。 从规模不变的关键点鲜明的形象特征。 IJCV， 2（ 60） :91 110， 2020。 [15]内藤吨，吨冢田，山田光，光 Kozuka，山本学，乐百氏车牌传递 underoutside环境的车辆识别方法。 电机及电子学工程师联合会 T，请设在 TECHNOL 49（ 6）： 2309 年至 2319 年 2020 年 11 月。 [16]内藤吨，吨佃，山田光，光 Kozuka。 牌照斜板鲁棒识别方法在不同的光照条件室外。 触发。 对 IEEE / IEEJ / JSAI国际智能交通系统，第 697702,1999。 [17] j 的耐荷，米布鲁日，光 Helmholt，学者 Pluim湖 Spaanenburg，河 Venema，米 Westenberg。 车牌与神经网络和模糊逻辑车牌识别。 诉讼的 IEEE 神经网络，珀斯， 西澳大利亚州，民国 21852903 国际会议。 1995。 [18] Teleghani，全德弗雷塔斯，学者和。 一个推动粒子过滤器：多目标检测与跟踪， ECCV， 2020。 [19] Papageiou， T. Poggio 着。 一个可训练的目标检测系统：车辆在静态图像检测。 麻省理工学院人工智能备忘录， 1673 年（ CBCL 人备忘录 180）， 1999。 [20]三施洛瑟，学者 Reitberger 尔欣斯，自动在高分辨率汽车检测都市风光一种自适应 3Dmodel 基础。 触发。 电机及电子学工程师联合会 / ISPRS 研讨会 “及以上城市的遥感领域的数据融合 ”。 2020。 [21]阁下 Schneiderman，金出武雄吨。 一种三维物体检测的统计方法应用到脸和汽车。 电机及电子学工程师联合会 CVPR， 2020。 [22]五夏皮罗湾 Gluhchev。 跨国公司车牌识别系统：分割和分类。 触发。 ICPR 10514651， 2020。 [23] j， Tomasi，跟踪好等特点。 触发。 电机及电子学工程师联合会机密。 计算机视觉与模式识别（ CVPR94） [24] j Sivic，答 Zisserman。 视频谷歌： 一个文本检索方法中的对象的视频匹配。 触发。 ICCV， 2020。 [25]克沙利文。 ，光贝克，答沃勒尔，角阿特伍德，体育 Remagnino，基于模型的车辆检测和分类 采用正交近似。 图像与视觉计算。 15（ 8）， 649654。 [26]，米琼斯。 快速检测对象使用一个简单的功能增强级联。 计算机视觉和模式识别 承认， 2020 年。 CVPR 2020。 诉讼 2020 年 IEEE 计算机上，体积学会会议： 1， 814， 页 511518。 2020。 [27]Yanamura，米藤，西山四，米索加，谷阁下，阁 下佐治。 提取和使用 Hough 变换牌照板块匹配跟踪和表决。 电机及电子学工程师联合会 IV2020 智能车辆研讨会。 6。 皮斯卡塔韦，美国新泽西州，2020 年。 Recognizing Cars Louka Dlagnekov, Serge Belongie Department of Computer Science and Engineering University of California, San Diego, CA 92093 {ldlagnek, Abstract License Plate Recognition (LPR) is a fairly well explored problem and is already a ponent of several mercially operational systems. Many of these systems, however, require sophisticated video capture hardware possibly bined with infrared strobe lights or exploit the large size oflicense plates in certain geographical regions and the (artificially) high discriminability of characters. In this paper,we describe an LPR system that achieves a high recognition rate without the need for a high quality video signalfrom expensive hardware. We also explore the problem of car make and model recognition for purposes of searchingsurveillance video archives for a partial license plate number bined with some visual description of a car. Our proposed methods will provide valuable situational information for law enforcement units in a variety of civil infrastructures. 1 Introduction License plate recognition (LPR) is widely regarded to be a solved problem with many systems already in operation. Some wellknown settings are the London Congestion Charge program in Central London, border patrol duties by the . Customs, and toll road enforcement in parts of Canada and the United States. Although few details are released to the public about the accuracy of mercially deployed LPR systems, it is known that they work well under controlled conditions. However, they have two main disadvantages which we address in this paper. Firstly, they require highresolution and sometimes specialized imaging hardware. Most of the academic research in this area also requires highresolution images or relies on geographicallyspecific license plates and takes advantage of the large spacing between characters in those regions and even the special character features of monly misread characters. Secondly, LPR systems by their nature treat license plates as cars’ fingerprints. In other words, they determine a vehicle’s identity based solely on the plate attached to it. One can imagine, however, a circumstance where two plates from pletely different make and model cars are swapped with malicious intent, in which case these systems would not find a problem. We as humans are also not very good at reading cars’ license plates unless they are quite near us, nor are we very good at remembering all the characters. However, we are good at identifying and remembering the appearance of cars, and therefore their makes and models, even when they are speeding away from us. In fact, the first bit of information Amber Alert signs show is the car’s make and model and only then its license plate number, sometimes not even a plete number. Therefore, given the description of a car and a partial license plate number, the authorities should be able to query their surveillance systems for similar vehicles and retrieve a timestamp of when that vehicle was last seen along with archived video data for that time. In this paper, we describe an LPR method that performs well without the need for expensive imaging hardware and also explore car make and model recognition (MMR). Because of the plementary nature of license plate and make and model information, the use of MMR can not only boost the LPR accuracy, but allow for a more robust car surveillance system. PreviousWork Most LPR systems employ detection methods such as corner template matching [11] and Hough transforms [12] [27] bined with various histogrambased methods. Kim et al. [13] take advantage of the color and texture of Korean license plates (white characters on green background, for instance) and train a Support Vector Machine (SVM) to perform detection. Their license plate images range in size from 79 38 to 390 185 pixels, and they report processing lowresolution input images (320 240) in over 12 seconds on a Pentium3 800MHz, with a % detection rate and a % false positive rate. Simpler methods, such as adaptive binarization of an entire input image followed by character localization, also appear to work as shown by Naito et al. [15] and [3], but are used in settings with little background clutter and are most likely not very robust. The most mon custom OCR approach used by existing LPR systems is correlationbased template matching [16], sometimes done on a group of characters [6]. Sometimes, the correlation is done with principal pone。