基于信息融合的道路前方车辆识别研究

将光学传感器与红外传感器进行信息融合以识别道路前方车辆.首先根据光学图像信息进行保守识别以初步确定车辆目标.在对相应的红外图像进行通道处理的基础上,充分利用其温度场信息提取车辆目标特征以构建车辆验证函数.以车辆验证函数值为依据建立基于最小风险的贝叶斯决策分类器,对光学初识别中得到的车辆目标进行验证,从而成功实现了光学图像与红外图像的信息融合.试验表明该方法能够利用红外热图信息在光学图像保守识别的基础上剔除误判目标,最终得到较为准确的识别结果.

Study on Identifying Vehicles on Road Frontage Based on Information Fusion

The information of optical sensor and infrared sensor was fused to identify vehicles on road frontage.First,the vehicles were restricted initially by conservative identifying according to the information of optical image.The corresponding infrared image was divided into three channels,and its temperature field information was used to summarize the characters of vehicles and construct checkout function.Bayes classifier based on minimal risk was constructed by the value of checkout function to validate the ob...