基于机器视觉的板材表面缺陷自动检测方法研究

Research on Automatic Detection Method of Plate Surface Defects Based on Machine Vision

摘要: 木质板材的表面缺陷不仅直接影响该产品的外观和质量，而且是影响板材分等级的重要因素之一。木质板材的表面质量是评估板材质量的重要指标，同时也能反映加工方法的合理性。由于板材表面缺陷种类有多种，同类缺陷在大小和外观形态上也各有差异，这种检测方式，一方面容易受人为因素影响，无法避免错检、漏检等情况发生，从而影响板材的质量；另一方面，浪费了大量的人力、财力，提高了成本，降低了竞争优势，还浪费了宝贵的林木资源。本文提出利用工业机器视觉语言对板材表面缺陷自动检测方法。该检测方法采用一种新的改良算法,解决了板材图像的配准问题；通过综合人工神经网络、模糊技术和遗传算法，建立了一种混合智能模型，利用人工神经网络技术，实现板材的表面缺陷检测。同时，设计出一种自动检测板材表面缺陷装置，该装置还可以按产品等级进行分拣。这样大大降低生产成本，减少产品检测过程的人为干扰因素，实现产品生产高度自动化，提高产品质量，能够产生很好的社会和经济效益。

Abstract: The surface defect of wood sheet not only directly affects the appearance and quality of the product, but also is one of the important factors affecting the classification of wood sheet. The surface quality of wood sheets is an important index for evaluating the quality of wood sheets, and it can also reflect the rationality of processing methods. Because there are many kinds of surface defects, the size and appearance of the same kind of defects are also different. On the one hand, this detection method is easy to be affected by human factors, which cannot avoid the occurrence of false detection and missed detection, thus affecting the quality of sheet metal; on the other hand, it wastes a lot of manpower and financial resources and improves the quality of sheet metal. This reduces the competitive advantage and wastes valuable forest resources. In this paper, an automatic inspection method for surface defects of sheet metal using industrial machine vision language is proposed. This method uses a new improved algorithm to solve the registration problem of sheet metal image. A hybrid intelligent model is established by synthesizing artificial neural network, fuzzy technology and genetic algorithm, and the surface defect detection of sheet metal is realized by using artificial neural network technology. At the same time, an automatic detection device for surface defects of sheet metal is designed, which can also be sorted according to product grade. This will greatly reduce production costs, reduce the human interference factors in product testing process, achieve a high degree of automation in product production, improve product quality, and can produce good social and economic benefits.