基于空间一致性的复杂场景钢坯端面字符定位

复杂场景中的字符目标定位是进行字符目标提取与识别的关键步骤，字符目标定位的准确程度决定着后续字符识别的正确与否。本文提出了一种基于空间一致性特征的复杂场景钢坯字符目标定位新方法。先对复杂场景钢坯图像进行分割，跟踪并标记分割图中每一个连通区域，然后根据钢坯端面字符在空间位置上排列的一致性，提取连通区域目标的质心点特征，并寻找排列成一条线的质心点，同时比较它们之间的区域面积及相互距离等特征，快速定位目标字符。对多幅实际复杂场景钢坯端面字符图像进行处理，结果表明我们提出的方法具有可行性和实用性。

The location method for steel billet character in complex environment based on spatial uniform feature

The location of characters object is the key step for characters extraction and recognition in complex environment, the validity of characters recognition is determined by object characters location. A novel method based on spatial uniform feature extraction for steel billet character location in complex environment is proposed. First we segment the steel billet image, track and label each union region in the segmented image, then extract the centroid feature according to the spatial uniform feature of the steel billet characters, search the centroid points arrayed in one line, and compare the area dimension and the distance of each union region in one line, thus locate the object character quickly and correctly. The processed results of the steel billet characters image demonstrate the feasibility and practicability of the proposed method.