Online Visual Inspection Method Research Of Ice Cream Stick Surface Defect

Ice cream stick is birch as raw materials,through multiple ways and production technology,and it is provided to the ice cream and other cold drinks.China is the largest producer of ice cream stick with huge volume.Due to the raw material of ice cream is not strict in the process of machining,ice cream stick will appear surface defects,such as black knot,mineral line,bark,pollution,split and color defects.At present,manufacturer mainly detect surface defects to finish quality sorting basing on the human eyes,which need a lot of workers next to the production line of ice cream stick to select defects one by one.This kind of detection is inefficient and grading subjective,and selecting quality is not guaranteed.Therefore,the research of automatic detection method for ice cream stick surface defects has the vital significance.Based on the above problem,this paper designes a set of complete inspection system software of the ice cream strick surface defects,and the corresponding prototype is developed.Based on the principle of machine vision method,realized the defects of online real-time detection.This paper briefly describes the hardware and software design of the ice cream stick defects detection system,analyses the characteristic of six kinds of defects on the ice cream strick surface.The defects are classified according to the analysis,the first kind of defects is small area defects with black knot,mineral line and pollution defects;The second category is large area defect with bark defect;The third type of defects is color defects,the fourth class of defects is split.Then corresponding detection schemes are designed according to the different characteristics of each type of defects,which achieve the quality classification of the ice cream stick surface defects.According to the latest quality testing standard of the ice cream stick manufacturer,this paper choose suitable cameras,lens,light source,computer and so on.Improved and developed the prototype system,maked field test and validated of the collection images for the software algorithm designed,which has achieved the testing requirements of enterprise standard.In this paper,the detection method designed is mainly to use the image gray value features and gaussian line detection to extract defect,image processing and detection classification,which has greatly solved the time-consuming problem of the complex identification algorithm.And using multi-thread parallel work also effectively reduces the software running time.All in all,the on-line detection system can effectively avoid the traditional artificial ice cream stick surface detection of defects,and realize the high efficiency,high speed online real-time detection,detection speed can reach 10 ice cream sticks per second and detection recognition rate can reach 99.42%,which has a certain practical application value,which has laid a good foundation for the development of ice cream stick online automatic quality testing.