基于彩色图像不变矩的交通标志特征提取方法

Traffic Signs Feature Extraction Method Based on Color Image Invariant Moments

本文提出了一种利用RGB颜色差分矩阵对Hu不变矩进行加权的新方法.并将该方法和Hu不变矩投影变换快速算法相结合,对彩色交通标志的特征值进行提取.实验表明,该方法可以在Hu不变矩不受目标图像大小、位置、旋转等因素影响的基础上,进一步减小光照条件及天气因素对彩色目标图像的影响,快速、准确地提取出目标特征值,为目标识别提供依据.

A new approach to weighting Hu invariant moments, which is based on the RGB color difference matrix, is proposed in this paper. By using the associated method of fast algorithm for Hu invariant moments and the weighting algorithm, the features of colorized traffic signs have been extracted. Experiments date for this method showed that the features extracted in this way not only were unacted on the image size, position, turn, but also decreased the influence of sunshine condition and weather factors. And the features of the target could be extracted exactly and fleetly. It Offered a way of target recognition.