| Mold is one of the most important equipments in injection molding industry, and its performance directly affects the quality of plastic products. What’s more, the mold takes a great part of molding corporates’ cost. Therefore, how to protect mold by modern technology is an important issue related to cost reduction and efficiency improvement. Mold monitoring systems provide an effective solution for such a problem.However, on the one hand, due to adopting the industrial computer as the main controller, the traditional monitoring system has high cost but low expansibility; on the other hand, there are two key technical problems:the problem of false detection casued by the mechanical vibration and the illumination change. From this view point, some key issues on the mold monitoring system are investigated deeply in this thesis. The main contents and achievements of this whole reseach are introduced as follows.To improve mold monitoring system’s expansibility and design flexibility, this dissertation builds the hardware platform with a core of TMS320DM6437, based on analyzing the hardware structure and the working prosess of the traditional mold monitoring system.To solve the problem of false detection casued by the mechanical vibration, a FAST-9feather-based image registration algorithm is proposed. The algorithm contains several main steps including the detection of FAST-9features, the establishment of16-D descriptors, the feature matching by the nearest neighbor method and the fitting of transformation parameters by the least square method. The experimental results show that this algorithm has low computational complexity and good robustness to illumination, rotation, and geometry changes, which makes it suitable for the mold monitoring system.To achieve a rapid detection of the abnormal information, a basic abnormality detection algorithm is proposed; to solve the false detection problem caused by the changing illumination, the LBP (Local Binary Patern) features-based abnormality detection algorithm is proposed. The experimental results show this algorithm can get accurate detection results when the illumination changes.For the realization of the TMS320DM6437-based mold monitoring system, firstly, this dissertation describes the key modules’characteristics of DM6437and its software design in detail; secondly, it designs a friendly touch screen interface based on On-Screem Display (OSD) module and the ZLG/GUI library; moreover, the transplantation of the algorithms and the optimization of software are discussed.To sum up, these works provide the necessary theoretical basis, experimental data and valuable research experience for mold monitoring systems. This is a research work of positive academic and practical significance. |
