外文資料翻譯---車牌識(shí)別-其他專業(yè)(參考版)
2025-01-23 10:24本頁面
【正文】 detection being an important requisite. We approached the license plate detection problem as a text extraction problem [5]. The detection method can be described as follows. A window of interest, of roughly the dimensions of a license plate image, is placed over each frame of the video stream and its image contents are passed as input to a classifier whose output is 1 if the window appears to contain a license plate and 0 otherwise. The window is then placed over all possible locations in the frame and candidate license plate locations are recorded for which the classifier outputs a1. In reality, this classifier, which we shall call a strong classifier, weighs the decisions of many weak classifiers, each specialized for a different feature of license plates, thereby making a much more accurate decision. This strong classifier is trained using the AdaBoost algorithm, and the weak classifiers are considered weak since they only need be over 50% accurate. Over several rounds, AdaBoost selects the best performing weak classifier from a set of weak classifiers, each acting on a single feature. 3 Make and Model Recognition As with the license plate recognition problem, detecting the car is the first step to performing make and model recognition (MMR). To this end, one can apply a motion segmentation method to estimate a region of interest (ROI) containing the car. Instead, we decided to use the location of detected license plates as an indication of the presence and location of a car in the video stream and to crop an ROI of the car for recognition. This method would also be useful for make and model recognition in static images, where the segmentation problem is more difficult. Character Recognition It was our initial intent to apply a binarization algorithm, such as a modified version of Niblack’s algorithm as used by Chen and Yuille [5], on the extracted license plate images from our detector, and then use the binarized image as input to a mercial OCR package. We found, however, that even at a resolution of 104 31 the OCR packages we experimented with yielded very poor results. Perhaps this should not e as a surprise considering the many custom OCR solutions used in existing LPR systems. Unless text to be read is in handwritten form, it is mon for OCR software to segment the characters and then perform recognition on the segmented image. The simplest methods for segmentation usually involve the projection of row and column pixels and placing divisions at local minima of the projection functions. In our data, the resolution is too low to segment characters reliably in this fashion, and we therefore decided to apply simple template matching instead, which can simultaneously find both the location of characters and their identity. The algorithm can be described as follows. For each example of each character, we search all possible offsets of the template image in the license plate image and recor
黨政相關(guān)相關(guān)推薦
鄂ICP備17016276號(hào)-1