現今大學內普遍設有化學實驗室,並備有許多不同種類與性質的化學藥劑,如何有效率管理化學藥劑,是一個非常重要的議題。 本研究目的基於聯合國主導推行的(Globally Harmonized System of Classification and Labeling of Chemicals, GHS)化學品分類與標示全球調和系統,建置一個使用主動式PTZ IP攝影機,依據GHS所定義之九種危害圖示,運用圖示顏色特徵技術搭配高斯模型(Gaussian Model)與形狀特徵七個不變矩(Moment Invariant)進行GHS危害圖示偵測,並針對偵測到的危害圖示利用貝氏分類器(Bayesian Classifier)進行9種危害圖示的辨識。本系統利用PTZ IP攝影機之光學變焦功能擷取影像進行分析,並利用GHS圖示特徵進行GHS圖示偵測與辨識,達成化學實驗室藥劑管理的目標。
Nowadays, universities have chemical laboratories equipped with various types and properties of the chemicals. Sometimes accident happened because of human negligence to place chemicals at unsafe position. Therefore, efficient chemical management becomes a very important issue. The purpose of this study is based on the Globally Harmonized System (GHS) for classification and label of chemicals introduced by the United Nation to build an automated PTZ IP camera, which is defined by GHS (9 kinds of hazard icons) to make use of shape features of Moment Invariant and color features of Gaussian Model technology to indicate hazard icons detection of the GHS, and using the shell type sorter (Bayesian Classifier) to recognize the hazard icons. This system utilizes optics zoom of the IP camera and the uniqueness of the GHS icons to detect, and recognize the images for analysis. Moreover, utilizing the medicament bottle features to judge the fall of the medicament bottles. Therefore, this research achieves the goal of the chemical laboratory medicament management.