中文摘要：

面向墙体裂缝检测技术难题,提出适合于实际工程应用的改进张量投票算法。针对现有方法在实际工程应用中易受噪声影响、精度与效率较低等难题,结合墙体裂缝的自身特征,提出3种改进的张量投票墙体裂缝检测算法,包括分块有选择性张量投票算法、竖向张量投票算法、轮廓膨胀过滤张量投票算法。以实际工程应用中某核电站安全壳壳壁的墙体裂缝检测对提出算法进行实验验证,并与原始张量投票算法以及现有裂缝检测方法进行比较分析。实验结果表明,改进后的算法可有效抑制噪声的影响,在精度和效率方面均优于原始张量投票算法和现有方法,且轮廓膨胀过滤张量投票算法效果最佳,适用于实际工程应用。

英文摘要：

This paper proposes an improved tensor voting algorithm to solve the key technical problem of the wall crack detection in the practical engineering applications.Aiming at the difficulties which are easily affected by the noise,and the relative lower efficiency and accuracy of the existing methods,we propose three improved tensor voting algorithms after giving a comprehensive consideration to the characteristics of the wall cracks.They are the improved tensor voting algorithm based on block selective strategy,the improved tensor voting algorithm based on vertical direction of the wall cracks,and the improved tensor voting algorithm based on outline expansion and filtering.In the experiments,the wall crack detection of a nuclear power station containment in the practical engineering application are implemented to verify the effect of the proposed methods.In addition,the original tensor voting algorithm and the existing crack detection algorithm are applied for comparison.Experimental results show that the three proposed methods can all suppress the noise,and their efficiency and accuracy are both better than the original tensor voting algorithm and the existing crack detection method.Especially,the proposed improved tensor voting algorithm based on outline expansion and filtering is the best,and it is suitable for the practical engineering application.