Research On Defects Detection Of Surface In Ice Cream Bar

Ice cream bar is provided to the ice cream and other cold drinks, it is Betula as raw materials,through multi-channel process. Due to the natural environmental factors and human factors, the surface of the ice cream bar may be black knot, mineral line, bark, pollution, such as color defects.At present, manufacturers mainly detect surface defects of ice cream bar basing on the visual. The artificial visual method has a lot of disadvantages. Not only can guarantee the quality, but also low efficiency, high labor costs and so on. Therefore, aiming at the actual situation of ice cream bar defects detection, this paper designed a set of ice cream bar surface defects based on machine vision visual on-line detection system.Firstly, the system selects the appropriate parameters of equipment for the construction of the hardware platform. According to the characteristics of gray histogram analysis, surface defects to be detected can be divided into three categories. In view of the characteristics of all kinds of defects, it respectively formulate different detection scheme. The first kind of defects is using fixed threshold binarization method to extract defect directly; The second flaw can be determined through the maximum connected region of grey value to judge whether there is this kind of defects; The third type of defects is to use an adaptive local threshold method and gray level histogram method to detect respectively. To some defects, such as the bark, it is used for edge detection algorithm for the detection of a second. Finally, using the ice cream bar quality classification standard to quality grading of detected flaws, it is calculated the ice cream bar final quality level, and real-time display on the screen.According to the design of the surface of the ice cream bar defects detection method, make the corresponding prototype. And through the scene test for many times, it can prove that the prototype of the indicators can reach enterprise standard of testing, including detection accuracy can be up to 98.34% on average. This paper’s detection methods in the light of the image grey value, so that we can solve the problem of complex recognition algorithm of time consuming. All in all, the system can effectively avoid the traditional artificial ice cream bar surface flaw detection of defects, so as to achieve high efficiency, high speed of online real-time detection.