基于局部方差的安全玻璃碎片自动检测算法

针对安全玻璃质量检验中碎片状态试验结果的判定问题,在分析了自然光线下获取的安全玻璃碎片图像特点后,提出了一种安全玻璃碎片自动检测与识别算法。首先,用基于局部方差的直方图均衡化算法进行目标增强;然后,用固定阈值进行图像分割,对缝隙线和碎片内部的噪声分别去除;最后用分水岭变换进行碎片识别。该算法能够准确刻画碎片的缝隙线,得到碎片的块数,并能够为安全玻璃碎片状态试验结果判定提供依据。

A Local Variance-Based Algorithm for Testing Safety Glass Fragmentation

This paper investigated the automatic test of safety glass fragmentation,an important part of the qualification of safety glasses.Based on an analysis of the characteristics of the fragmentation images obtained under natural lighting conditions,we proposed an algorithm for the automatic testing and recognition of glass fragments.The algorithm first used local variance-based histogram equalization to enhance the objects in images,and then partitioned the images by using a fixed threshold that removes the noises inside the fragmentation lines and fragments,and finally used the watershed transformation for fragment recognition.Experiments using images from real fragmentation tests showed that the algorithm could accurately determine the fragmentation lines and the number of fragments in the images,and could be used in standardized qualification tests of safety glass fragmentation.