算法报告—计算机多尺度辅助定位车牌算法（18页）.pptxVIP

计算机多尺度辅助定位车牌算法Multi-scale auxiliary positioning license plate through algorithm Abstract: There are the characters of large data， strong effectiveness andcomplex environment in aencroaching vehicles，the detection of illegal encroaching vehicles requires amount of experimental work and labor． In the process of positioning license plates， people cannot avoid some 摘 要: 违法占道拍摄出的单帧车辆图像具有数据量大、时效性 强，检测环境复杂等特点． 对其检测需要花费大量的人力与single-frame image of illegal物力． 并且人们在定位过程中，无法避免因经验、疲劳等方面的干扰，导致遗漏和错误定位． 为此本文从视 觉感知角度提出计算机多尺度辅助定位车牌算法． 模拟视觉感知原理，从车辆特征、纹理特征、颜色特征尺度，逐次聚 焦至车牌所在区域． We put forward a complete license plate positioning process for the single-frame image and propose an accurate location algorithm for license plate area based on boundary pair． The experiments on real traffic images show that the proposed algorithm has a higher accuracy rate when the image is acquired fromthe head or rear of vehicles， and is consistent with the 我们提出了完整的单帧图像车牌定位流程． 并且提出基于边界对的车牌区域准确定位算 法． 通过对实拍的交通图像实验，表明本算法对于正对的车辆有较高的准确率，符合人类视觉感知的过程可实时的对图像进行车牌 检测，可同时检测单幅图片的多个车牌． 但对于光线过暗、过强或者颜色失真的情况，仍需要进一步的研 究． The types of algorithms are built from a single featureperspective. they are only suitable for license plate loc in a simple environment.Texture featureColor feature The above algorithm only constructs the以上算法仅从图像特 model from the perspective of image 征角度构建模型，忽 features, ignoring the visual cognitio略n 视觉认知过程，难 process, and it is difficult to ensure t以he同时保证速度与精 speed and accuracy requirements at the度的要求same time. 程进行研究，提出计算算法，从车、纹理和颜色三尺度对图像进行层层聚焦，并提出边缘对准确的实现车牌定位，满足实际需求．Based on the research of visual ? 本文基于对视觉认知过cognition process, this paperproposes a computer multi-scaleauxiliary license plate detection机多尺度辅助车牌检测algorithm, which focuses on the image from three dimensions of car, texture and color, and proposes the edge-to-edgelicense plate location algorithm t的o 车牌定位算法，快速 realize the license plate locationquickly and accurately. 视觉感知车牌 计算机构建变形部分模型聚焦至 车辆所在区域( 车尺度)构建边缘对算子对疑似 车牌区域进行定位( 纹理尺度)根据车牌设计特点，从颜色组合角度确定车