[1]赵海燕,王 婧,刘晓宇*,等.doi: 10.3969/j.issn.1001-3849.2025.11.018复杂背景下合金钢电镀工件表面缺陷机器视觉挖掘[J].电镀与精饰,2025,(11):128-134.
　Zhao Haiyan,Wang Jing,Liu Xiaoyu*,et al.Machine vision mining of surface defects on alloy steel electroplated workpieces in complex backgrounds[J].Plating & Finishing,2025,(11):128-134.
点击复制

doi: 10.3969/j.issn.1001-3849.2025.11.018复杂背景下合金钢电镀工件表面缺陷机器视觉挖掘()

《电镀与精饰》[ISSN:1001-3849/CN:12-1096/TG]

卷:
期数:
2025年11
页码:
128-134
栏目:
出版日期:
2025-11-30

文章信息/Info

Title:
Machine vision mining of surface defects on alloy steel electroplated workpieces in complex backgrounds
作者:
赵海燕; 王 婧; 刘晓宇*; 刘 琨; 肖楷乐
(北京联合大学 应用科技学院,北京 100012)
Author(s):
Zhao Haiyan;  Wang Jing;  Liu Xiaoyu*;  Liu Kun;  Xiao Kaile
(College of Applied Science & Technology, Beijing Union University, Beijing 100012, China)
关键词:
复杂背景; 合金钢电镀工件; 表面缺陷; 机器视觉挖掘
Keywords:
complex background;  alloy steel electroplated workpieces;  surface defects;  machine vision mining
分类号:
TP391
文献标志码:
A
摘要:
合金钢电镀工件加工环境中，机械设备、传送带、夹具等元素会形成复杂背景，同时光照不均、尘埃、油污等因素也会干扰图像质量，导致工件图像缺陷边缘信息不显著，增加了前景与背景区域分割难度，直接进行工件缺陷识别难以准确区分缺陷区域，影响缺陷挖掘效果。针对这一问题，本文提出了一种复杂背景下合金钢电镀工件表面缺陷的机器视觉挖掘方法。通过引导滤波和直方图均衡化对合金钢电镀工件图像进行预处理，消除背景干扰并提高对比度，获取包含更多细节的细粒图像。采用拉普拉斯变换方法对细粒图像进行边缘增强处理，突出缺陷边缘信息，显著提升图像边缘清晰度和对比度。使用自适应阈值算法对增强后的图像进行前景与背景分割，提取仅包含工件及其缺陷的前景图像，减少数据处理复杂性。应用谱多流形聚类技术，将分类器中的相似点分配到不同的缺陷流形结构中，对缺陷区域进行精准划分，更准确地识别微小缺陷，实现复杂背景下合金钢电镀工件表面缺陷的机器视觉挖掘。通过实验证明，应用本文所提方法对图像边缘进行处理后，图像边缘清晰度为3.2 lp/mm，对比度方差为0.96，图像分割后的IoU值为99.6%，说明该方法在工件表面缺陷视觉挖掘中具有较好的效果。
Abstract:
In the processing environment of alloy steel electroplating workpieces, complex backgrounds are formed by mechanical equipment, conveyor belts, fixtures, and other elements. At the same time, uneven lighting, dust, oil stains, and other factors can also interfere with image quality, resulting in insignificant edge information of workpiece defects and increasing the difficulty of foreground and background area segmentation. Directly identifying workpiece defects is difficult to accurately distinguish defect areas and affects the effectiveness of defect mining. This article proposes a machine vision mining method for surface defects of alloy steel electroplated workpieces in complex backgrounds to address this issue. By guiding filtering and histogram equalization to preprocess the images of alloy steel electroplated workpieces, background interference is eliminated and contrast is improved to obtain fine-grained images with more details. The Laplace transform method is used to perform edge enhancement on fine-grained images, highlighting defect edge information and significantly improving image edge clarity and contrast. An adaptive thresholding algorithm is then applied to segment the foreground and background of the enhanced image, extracting foreground regions containing only the workpiece and its defects, thereby reducing data processing complexity. By applying spectral multi manifold clustering technology, similar points in the classifier are assigned to different defect manifold structures, accurately dividing the defect area and more accurately identifying small defects, achieving machine vision mining of surface defects on alloy steel electroplating workpieces under complex backgrounds. Through experiments, it has been proven that applying the method proposed in this article to process image edges results in an image edge clarity of 3.2 lp/mm, a contrast variance of 0.96, and an IoU value of 99.6% after image segmentation. This indicates that the method has good performance in visual mining of surface defects on workpieces

参考文献/References:

[1].陈启鹏, 熊巧巧, 黄海松, 等. 基于YOLOV4的工件表面质量在线检测方法研究[J]. 包装工程, 2023, 44(3): 148-156.
[2].章文显, 桑劲鹏, 胡珈源, 等. 等温淬火球墨铸铁厚壁工件无损检测方法的对比研究[J]. 铸造, 2024, 73(5): 683-687.
[3].孙贺斌, 周治伊, 吕岩婷, 等. 基于低频双探头超声波的盆式绝缘子缺陷检测方法研究[J]. 绝缘材料, 2023, 56(1): 110-114.
[4].邹宸玮, 么娆. 基于机器视觉的超声相控阵缺陷检测研究[J]. 计算机科学, 2023, 50(增2): 383-388.
[5].袁忠大, 刘傲天, 张小锋, 等. 基于机器视觉的飞机电镀部件曲表面无损检测[J]. 电镀与涂饰, 2024, 43(7): 43-50.
[6].Yi H A, Lv X, Shu A H, et al. Few-shot detection of surface roughness of workpieces processed by different machining techniques[J]. Measurement Science & Technology, 2024, 35(4): 045016.1-045016.12.
[7].张毅, 范玉刚. 基于深度学习与域自适应的工件涡流热成像的缺陷检测[J]. 红外技术, 2024, 46(3): 347-353.
[8].刘艳菊, 王秋霁, 张惠玉, 等. 基于改进SSD的工件表面缺陷检测[J]. 热加工工艺, 2024, 53(2): 134-139.
[9].张书赫, 曹良才. 基于数字衍射的单幅眼底图像增强[J]. 光学精密工程, 2024, 32(15): 2429-2438.
[10].韩晓微, 张云泽, 谢英红, 等. 动态异构特征融合的水下图像增强算法[J]. 控制与决策, 2023, 38(6): 1560-1568.
[11].郑浩君, 王振, 张佳鹏, 等. 基于混合注意力模块改进StarGAN的水下图像增强[J]. 南方水产科学, 2025, 21(1): 185-196.
[12].冯欣, 王思平, 张智先, 等. 一种面向机器视觉感知的暗光图像增强网络[J]. 计算机应用研究, 2024, 41(6): 1910-1915.
[13].陈华腾, 刘磊, 钱芸生, 等. 基于Retinex里双分量改进的微光图像增强方法[J]. 应用光学, 2024, 45(4): 819-827.
[14].常戬, 韩旭. 结合导向滤波与自适应算子的水下图像增强[J]. 计算机工程与应用, 2023, 59(4): 216-223.

更新日期/Last Update: 2025-11-20