中文摘要：

对车载红外图像进行具有空间立体感的彩色化处理可以改善车载红外图像的理解效果，更好地辅助驾驶．首先用基于遗传算法优化BP神经网络模型的单目红外图像深度估计方法获取单幅红外图像中的景物深度，然后用深度信息调制彩色化后的红外图像的饱和度，利用透视学中饱和度变化来区分表达景物的深度，得到视觉上有空间立体感的彩色车载红外图像．试验结果显示饱和度调制后的彩色车载红外图像具有视觉上的空间立体感，有助于环境感知和安全驾驶．

英文摘要：

Colorizing the vehicle infrared image with stereoscopic perception can help the driver understanding the vehicle infrared image better and assist driving safely. Firstly, a depth estimation method of monocular infrared image depth based on BP neural network and optimization by genetic algorithm was proposed. The depth of scene in an infrared image was obtained using the monocular infrared image depth estimation method based on BP neural network model and optimization by genetic algorithm. Based on the perspective theory, the depth of scenes in a color infrared image was discriminated using the saturation variation. Therefore, a color vehicle infrared image with stereoscopic perception was obtained. The experimental results show that the color vehicle infrared image after saturation modulation has stereoscopic perception which can assist environmental perception and driving safely.