基于特征匹配的集装箱识别与定位技术研究

集装箱识别与定位是实现港口自动化的关键技术之一。模板匹配是一种较常用的 图像识别方法，但传统的模板匹配算法仅基于像素的灰度信息进行对比匹配，当外部环境稍有 变化时，算法的鲁棒性会大幅下降，且存在匹配效率不高，适应性不强等问题。提出基于目标 颜色模板预定位与变步长图像块匹配分割相结合的算法，将具有形状不变性的特征向量作为集 装箱实时识别与定位的重要依据，实验结果显示该识别方法能有效提高图像分割效率，同时对 目标平移、旋转、尺度变化具有适应性。

Container Recognition and Location Technology Based on Feature Matching

Container recognition and location is one of the key technologies to realize port automation. Template matching is one of the basic and frequently used image recognition methods. Traditional template matching algorithm only based on gray information of pixels to compare and match objects. When external environment changes slightly, the robustness of the algorithm is greatly reduced. Moreover, there are some problems, such as low matching efficiency and adaptability, etc. So, an algorithm was presented which combined with color template pre-position and image segmentation based variable step size block matching. The feature vector with shape invariance was extracted, which is an important basis of the container real-time recognition and location. The result shows that this recognition method not only improved image segmentation efficiency, but has great adaptability for target translation, rotation and scale change.