CAJ | 학술논문

针对电视台标所处背景的复杂性，现有的台标识别方法尚不能对各种台标进行有效识别，提出了一种基于统计分类的自动台标识别方法，采取先分类后识别等多种有效措施，可对复杂背景下的台标进行正确识别。通过对大量视频数据的验证证明，该方法识别精度高，具有较强的适应性。
침대전시태표소처배경적복잡성，현유적태표식별방법상불능대각충태표진행유효식별，제출료일충기우통계분류적자동태표식별방법，채취선분류후식별등다충유효조시，가대복잡배경하적태표진행정학식별。통과대대량시빈수거적험증증명，해방법식별정도고，구유교강적괄응성。
TV station logo is an important sign,and the technology of automatic TV station logo recognition can be used for supporting video viewing behavior statistics.For the complexity of the TV station logo's background around, existing methods can not effectively identify every kind of TV station logo.This paper brings forward an automatic TV station logo recognition algorithm based on statistical classification.Through recognition after classification and series of effective measures,the algorithm can correctly identify the TV station logo under a variety of complex backgruund.By a large number of video data validation, it proves that the recogni- tion algorithm has high identification accuracy and strong adaptability.