基于纹理主、旁瓣特征的雪糕棒裂缝缺陷检测
发布时间：2018-11-17 10:55
【摘要】：裂缝是雪糕棒表面一种严重缺陷,对雪糕棒的加工和使用影响极大,然而部分又细又浅的裂缝与雪糕棒表面的木纤维纹络具有诸多相似之处,使得当前检测算法提取效果不佳。针对该问题,在对裂缝纹理及木纤维纹络的特征进行详细分析的基础之上,提出了一种基于纹理主瓣和旁瓣灰度特征相结合的检测方案。首先建立纹理主瓣和旁瓣灰度特征提取基本模型;然后提取雪糕棒表面头部全部纹理的边缘;接着,根据建立的模型提取上一步骤所得各边缘相应纹理的主瓣和旁瓣灰度特征量,并根据主瓣特征量的数值大小初步锁定其中属于裂缝纹理的候选边缘(其中包括全部的裂缝边缘和部分木纤维纹络边缘);最后,通过旁瓣特征量与预设阈值的数值关系识别出上一步骤候选边缘中的裂缝纹理边缘,从而实现裂缝缺陷的检测。在自建图库SUT-I3上进行了测试,结果显示所提方法在裂缝缺陷漏检率为0的前提下,其误检率低至6.07%,相对于其他雪糕棒或木材表面裂缝检测方法其误检率最少降低了9.29%,表明了所提方法的高效性,具有实际应用价值。
[Abstract]:Crack is a serious defect on the surface of ice cream rod, which has a great influence on the processing and use of ice cream rod. However, some thin and shallow cracks have many similarities with the wood fiber veins on the surface of ice cream rod, which makes the current detection algorithm poor. In order to solve this problem, based on the detailed analysis of crack texture and wood fiber texture features, a detection scheme based on grayscale features of texture main lobe and sidelobe is proposed. Firstly, the basic models of grayscale feature extraction of main and sidelobe are established, then the edge of all the texture on the surface of ice cream bar is extracted. Then, according to the established model, the grayscale features of the corresponding texture of each edge are extracted from the previous step. According to the value of the characteristic quantity of the main lobe, the candidate edges which belong to the fissure texture are preliminarily locked (including all the fissures and some wood fiber fringes). Finally, through the numerical relationship between the sidelobe characteristic quantity and the preset threshold value, the edge of the crack texture in the candidate edge of the previous step is identified, and the crack defect detection is realized. The test results on SUT-I3 show that the error detection rate of the proposed method is as low as 6.07 when the crack defect leakage rate is 0. Compared with other ice cream bars or wood surface crack detection methods, the false detection rate is reduced by 9.29%, which indicates that the proposed method is effective and has practical application value.
【作者单位】： 沈阳工业大学视觉检测技术研究所;辽宁省机器视觉重点实验室;
【基金】：国家自然科学基金(61271325)项目资助
【分类号】：TP391.41;TS277
，
本文编号：2337524
[Abstract]:Crack is a serious defect on the surface of ice cream rod, which has a great influence on the processing and use of ice cream rod. However, some thin and shallow cracks have many similarities with the wood fiber veins on the surface of ice cream rod, which makes the current detection algorithm poor. In order to solve this problem, based on the detailed analysis of crack texture and wood fiber texture features, a detection scheme based on grayscale features of texture main lobe and sidelobe is proposed. Firstly, the basic models of grayscale feature extraction of main and sidelobe are established, then the edge of all the texture on the surface of ice cream bar is extracted. Then, according to the established model, the grayscale features of the corresponding texture of each edge are extracted from the previous step. According to the value of the characteristic quantity of the main lobe, the candidate edges which belong to the fissure texture are preliminarily locked (including all the fissures and some wood fiber fringes). Finally, through the numerical relationship between the sidelobe characteristic quantity and the preset threshold value, the edge of the crack texture in the candidate edge of the previous step is identified, and the crack defect detection is realized. The test results on SUT-I3 show that the error detection rate of the proposed method is as low as 6.07 when the crack defect leakage rate is 0. Compared with other ice cream bars or wood surface crack detection methods, the false detection rate is reduced by 9.29%, which indicates that the proposed method is effective and has practical application value.
【作者单位】： 沈阳工业大学视觉检测技术研究所;辽宁省机器视觉重点实验室;
【基金】：国家自然科学基金(61271325)项目资助
【分类号】：TP391.41;TS277
，
本文编号：2337524
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2337524.html
