Research On Quality Inspection Technology For Health Wine Product Based On Machine Vision

With the growing of China's health wine market and the improvement of quality standards of products, the traditional manual detection methods can not meet the demand of high-speed production line. Inspection equipment based on machine vision can quickly obtain real-time image information of products, accurately judge the quality of the product and greatly improve production line automation. Through researching on the key technology of visual inspection equipment, this paper designed a quality inspection system based on machine vision for health wine product, described in detail the development process and the detection algorithm.Firstly, this paper introduced the research background and significance, analyzed the development situation in China's health wine market and the current research on visual inspection equipment for health wine products both at home and abroad, and demonstrated the essentiality of development on visual inspection equipment for China's standards and production environment, and then this paper illustrated the structure of Machine Vision System, explained the applications, discussed the visual inspection technology research and development trends, and elaborated the key technology of Visual imaging system including optical lens, video, image acquisition card and etc.According to health wine production line detection requirements, this paper put forward a visual detection system program. In mechanical respect, an overturn control plant was designed to stir foreign body sinking in the bottom. In order to acquire each bottle sequence images without deviation angle, used a tracking mirror. In control system respect, designed two control programs based on PC+motion control card+ data I/O card and based on PC+PLC according to the difference of movement control mode and control module. The paper also introduced the software system, analyzed the composition of the software system, working principle and the module function.Preprocessing images is an important step in visual inspection. After analyzing the advantages and disadvantages of gray stretch and histogram equalization, this paper studied the histogram specification image enhancement method. Against traditional noise suppression methods deficiencies, this paper proposed a method of noise suppress by selecting the optimal masks to lay the foundation for the next image processing, which not only effectively suppressed image noise but also retained the image edge details well.Aim at the quality problems existing in health wine products such as foreign, bottle defect, insufficient filling, this paper designed corresponding detection algorithms. It was vital for foreign detection to analyze connectivity of each target pixel in sequences image, compute the invariant features, use invariant features search similarity target in next frame and judge whether foreign body existed according to the change of matched target y-coordinate. In accordance with the feature of the bottle image, utilized the gradient edge detection operator for bottle defect detection, and then removed the false edge so as to improve detection accuracy through analyzing the pixel value in the area which is surrounded by the edge. This paper studied algorithms based on the Hough transform for level detection. Extracted preliminary level by making use of the mask processing algorithm to reduce computation and improve the detection speed well.Finally, the paper summarizes the work and achievements, and the prospect in future work.