基于模糊形状判别的鲁棒交通标志检测算法 robust traffic sign detection using fuzzy shape recognizer.pdfVIP

I匝l_圜l圃 Research．Development 基于模糊形状判别的鲁棒交通标志检测算法木 RobustTrsmcDetection Sign UsingFuzzyShapeRecognizer 丁淑艳李伦波潘海朗 (南京理工大学电子工程与光电技术学院，南京210094) 摘要：针对交通标志存在尺度、旋转、倾斜、表面被污损或部分被遮挡等退化情况，提出一种基于模糊形状 判别的鲁棒交通标志检测算法。该算法在HSV彩色空间进行颜色分割的基础上，提取目标区域的对称局部特 征，根据设计的模糊形状判别算法来判定目标区域的形状，进而检测出交通标志。采用该检测算法对不同天气情 况、不同道路场景下的3000多幅自然场景图像进行交通标志检测实验，结果表明该检测算法不仅具有较高的检 测率，而且具有良好的鲁棒性。 关键词：辅助驾驶系统交通标志检测HSV彩色空间模糊形状判别 DOI：10．3969／j．issn．1007—080x．2012．12．001 noveltraffic basedon for Abstract：A detection is thedetectionof sign algorithm fuzzyshaperecognizerproposed traffic innatural sufferfromvariousformsof as signs environments，whichusually imagedegradation，suchscaling， RGB isconvertedintoHSVcolor occlusion．Every segmentedby rotation，tilt，damage，andpartial image space，and thehueandsaturationthresholds．Thelocalfeaturesofthe of extracteda interest(ROI)are regions using the ofROIisdetermineda than3000road werecollected detector，and shape by fuzzyshaperecognizer．More images underdifferentweatherconditionsanddifferentroad usedfor this results scenes，andtesting showthatthe not obtainsdetection alsohas robust proposedalgorithmonly high accuracy，butgood performance． assistance traffic detectionHSVcolor Keywords：drivingsystems sign spacefuzzyshaperecognizer 0引 言