提出一种车载红外视频快速彩色化方法，利用轮廓特征点跟踪获取每帧物体类别的轮廓区域，采用类别特征色彩对各区域传递色彩。构建各景物样本特征色彩集，以各类景物在自然彩色图像中表现出来的特征色彩作为红外图像中对应景物的色彩；利用改进的高效K-Means方法对红外关键帧进行聚类，得到分割区域，提取轮廓特征点；通过KLT算法跟踪特征点，得到其在下一帧中的位置并同时修正，采用B样条插值进行轮廓复原，得到该帧的各类别轮廓区域；最后将特征色彩按类别赋予各区域，从而给各帧图像着上合适的颜色，实现红外视频序列的快速彩色化。实验结果表明, 该方法与基于运动估计的算法相比可提高近5倍的处理速度，并且能够得到与自然景物色彩较接近的彩色化视觉效果。

A fast algorithm of vehicle infrared video colorization was proposed. Contour feature points tracking was utilized to get the contours and areas of different object classes in each frame. Characteristic colors of different classes were transferred to the corresponding areas in each frame. Firstly, we constructed the characteristic color sets of different classes of scenes. The characteristic colors of different scenes in the natural color images were extracted as the corresponding scenes' colors in the infrared images. Secondly, we clustered and segment the scenes in the key infrared video frame using the improved efficient K-Means clustering method, and the contour feature points were extracted. Thirdly, we tracked the contour feature points using the Kanade-Lucas-Tomasi (KLT) algorithm to get and correct their positions in the next frame. The B-spline interpolation method was used to reconstruct the contours. The contours and areas of different object classes in each frame were obtained. Finally, characteristic colors of different classes were transferred to the corresponding areas in each frame. Infrared video colorization was realized and the scenes were with reasonable colors. Experimental results show that the processing speed of the proposed method is nearly 5 times as fast as that of the algorithm based on motion estimation and it can give a natural color look to the infrared video.