一种基于纹理的车辆图像分割与提取算法

Texture-based Vehicle Image Segmentation and Extraction Algorithm

提出一种基于纹理特征的车辆分割与提取算法,该算法充分利用了车辆纹理、亮度、色度的重要信息,通过循环迭代法选取最优的阀值,通过计算纹理掩码、亮度掩码、色度掩码,经过形态学操作和边缘提取,最终实现车辆分割。实验结果表明,本文的算法在多种复杂情况下具有较好的效果。

A texture feature based vehicle image detection algorithm is proposed in this paper, which considers about the important information of vehicle texture, luminance and chrominance. Based on iterative selecting optimal thresholds, texture mask, luminance mask and chrominance mask are computed, and morphological operations are performed, so as to obtain a vehicle contour and realize vehicle segmentation. Experimental results show that the proposed method is good enough. It could segment and extract vehicle exactly from a complex background scene.