中文摘要：

采用基于视觉检测的乘员分类识别算法，将副驾驶位上的乘员按照空座、成人、儿童和婴儿4种类型进行分类识别。该算法利用勒让德（Legendre）矩从检测的乘员图像边缘信息中提取出乘员特征，再利用支持向量机方法对乘员进行分类。对600张照片进行了分类识别试验的结果表明，准确率达到99％以上。

英文摘要：

A visual detection-based occupant classification/recognition algorithm is adopted to recognize and classify the empty, adult, child and baby four types of occupants at co-driver seat. The algorithm uses Legendre moments to extract occupant features from detected edges of occupant images, and utilizes support vector machine technique to conduct occupant classification. The results of recognition/classification test on 600 photos show that its accuracy reaches 99%.