【作者】 李辉；

【导师】 赵一丁；

【作者基本信息】 东北大学 ， 控制理论控制工程， 2012， 硕士

【摘要】 常用的食品异物检测方法有可见光检测、近红外检测、磁共振成像检测、超声波成像检测和X射线成像检测六种方法,本文采用X射线数字成像技术作为食品异物的检测手段。它的应用范围广,可以对包装食品、禽畜肉类以及散包食品中混入的石头、金属、玻璃、橡胶等类型的异物进行检测,其实用性强,可进行实时性的在线检测,相比于其他的检测方法具有明显的技术优势。本文以袋装瓜子为研究对象,使用X射线数字检测技术,对不可见的包装袋内的异物进行检测识别。将三种常见的异物混入袋装瓜子中模拟瓜子加工生产过程中异物混入的情景,使用手提式X射线透视仪对其进行照射,并使用USB图像采集棒将得到的X射线数字图像传输到计算机,以进行X射线瓜子异物图像的处理和识别。X射线瓜子异物图像预处理是后续图像处理的基础,本文分析了X射线瓜子异物图像噪声的来源并采取了相应的噪声消除；针对X射线瓜子异物图像对比度低等问题进行了图像的增强处理,经过去噪和增强后的图像能更准确的反应异物的信息。在X射线瓜子异物图像特征提取过程中,边缘提取是进行异物识别的有效方法。本文分析了几种经典边缘提取方法,以及基于形态学梯度的X射线瓜子异物图像的边缘检测方法。对与瓜子密度差异较大的异物,采用了行灰度曲线的特征提取法,根据含异物图像异物所在处和不含异物图像同一行灰度值的显著差异,可以判定出异物的存在。而对于密度与瓜子较为相近的异物,我们选择了图像减影的方法进行判别。我们分别使用了自适应中值滤波和自适应形态学法模拟背景图像,将含异物的图像与模拟的背景图像进行减影操作,从而得到异物目标图像。比较得到的异物目标图像,发现形态学模拟背景图像的方法更能清晰的保留异物的边缘和轮廓,获得了理想的实验效果。本文将X射线检测技术应用于袋装瓜子的异物检测,能够成功地识别出瓜子中的异物,这对同类食品生产企业的食品异物检测工作具有重要的指导意义。更多还原

【Abstract】 The common detective methods for foreign bodies in food include visible light, near-infrared, magnetic resonance, ultrasonic wave imaging and x-ray imaging six kinds of detective methods. As it has a wide range of application in detection, we choose the x-ray imaging technology as the detective method for the foreign bodies detection in food. It can be used in the detection for foreign bodies in Packing food, Shrimp&Scallop and bulk food which mixed with stone, metal, glass, rubber and other types of foreign bodies. The x-ray imaging detective methods has obvious advantage than other detective methods in practicality and real-time online detectionwe choose the bagged sunflower seeds as the object of our study. We can detect and distinguish the invisible foreign bodies that exist in bags by using the x-ray digital detection technology.Three common foreign bodies are mixed with the bagged sunflower seeds for simulating scene that foreign bodies are mixed in the seeds when processing in factory. We use the portable x-ray instrument to get the x-ray digital image by the USB image acquisition card for further processing.The image preprocessing of foreign bodies is the basis of subsequent image processing. In this thesis we analyzes the source of the image noise that exist in the x-ray image and take the appropriate algorithm to undo noise; Image enhancement for low contrast of x-ray foreign bodies image is also need, we can get more accurate information from the images after denoising and enhancement processing.In the feature extraction process of x-ray foreign body image, image edge extraction is an effective method for foreign body identification, in this paper we analyze several classical edge detection methods, as well as the edge detection method based on morphological gradient. For the foreign body which has greater density than sunflower seeds, we adopt the line gamma curves feature extraction method. According to the significant differences of gray value on the same line where the foreign body lies in image containing foreign bodies and the one doesn’t have, we can judge the presence of the foreign body in the seeds. If the density of the foreign bodies is similar to the seeds’, we can choose the image subtraction method. The adaptive median filter and adaptive morphology method are chosen for simulating the background of the image. Subtract of the original image and the background image, we can get the target image of the foreign bodies. Compare the target images of the foreign bodies obtained by this two methods, we can get a satisfactory experiment result when we choose the adaptive morphology method in simulating the background of image.Apply the x-ray detection technology in the detection of foreign bodies of the bags of seeds, we can identify the foreign bodies in the seeds successfully. It has an important guiding significance for the similar food production enterprises in the detection of d foreign bodies.更多还原