mold cavity a case study of automated inspection -

by:Hengju     2019-06-27
This paper describes the application of machine vision technology, which can be used to automatically detect the placement of wrong labels during the injection molding process of manual loading.In this application, simple techniques involving counting features are used to check the direction of a particular direction.If the part is placed incorrectly, the visual system sends a signal through its input and output (I/O) system and activates external devices such as a buzzer or warning light.The specific application of more advanced technologies such as character recognition and template matching in checking print quality is also discussed.In order to divide the region into foreground and background, it is necessary to threshold the image.The threshold.T.Basically setting is considered the boundary between dark pixels.i.e.l(m.N) the global feature of the image exported from the histogram is used to modify a single pixel value.The comparison of Point operations is usually achieved by simple scaling.The approach is slightly different in global operations.A technique called histogram equalization is used to redistribute pixel values to produce a uniform histogram.In an image with m rows and n columns, the bit resolution is r.The ideal histogram will be consistent with the (m x n/2 sup r) pixels for each gray scale.In order to achieve the desired effect, the spatial distribution of pixels is usually changed.Examples of geometric operations include image amplification, rotation, and transformation.Typically, these operations involve mapping functions that convert a set of pixels on one location (x, y) to another (x \ ', y \').Frame-Image-based operations are basically performed or achieved with multiple images.An example of this kind of operation commonly used in 17 \ '18 checks is a point-by-A pixel is compared or subtracted from the point of another pixel.Then, depending on how the newly constructed image compares to the original or some reference pixel set, it can be passed as good or as bad.Another algorithm that can be used is an algorithm that compares the distance between similar features.This algorithm, called template matching, uses data from known images as a training set.The data is usually composed of the distance between the features that act as vectors and/or their similarity.The unknown image vector data is compared with the image vector data of the training set.Set the threshold to define the success level at which the unknown image is compared to the training data.This paper uses the method of template matching to identify the printing quality of plastic.Such as casting molds, mold manufacturing, plastic injection molds, etc.In order to obtain printing on plastic molding, most processes are pre-The printed plastic template to be placed in the mold cavity, as shown in Figure 1.Figure 2 shows an example of a complete molded part with a printed display.The template is inserted into the cavity, usually upside down, in order to obtain the correct direction of the print.After the template is fixed in the mold cavity, the mold is closed and the molding is completed on the print.Then retrieve the part and check the quality manually.Usually, this inspection process is not done for all parts, and sometimes, the tired operator may miss the parts that are found to be poorly printed before the parts are shipped to the customer.In order to avoid heavy penalties usually due to the supply of defective parts, an intelligent vision system can be used to ensure that the template is in the right direction before closing the half part of the mold.Figure 3 shows a schematic diagram of how to do this automatically.To test it, a prototype was developed with a DVT (TM) series 600 camera.Once the template is placed in half of the mold, the camera is set to capture the image.Next, many algorithms (or tools) are used to check that the print direction is correct.Two linear feature counting tools are used.One is set at the top of the left edge of the print and the other is in the lower left corner.Set lower tools to identify the presence of lines (in this case, dark features of the specified minimum thickness in pixels), while higher tools are set to identify any dark or light-colored featuresWith Framework (TM) software, 18 it is quite easy to apply and set up these tools to perform the required checks without having to perform complex image processing and analysis.The result of each function count is then sent digitally to the camera I/O board because it "passes" or "fails ".\ "With this software, it can be configured by switching through digital I/O settings.The signal of the camera board is then sent to the input of the programmable logic controller (PLC.If the template is inserted incorrectly, the PLC is programmed to turn off the sound alarm (BUZZER) and turn on the warning light.Figure 4 is an application example of the linear feature counting tool used to perform this check.During the setup of this test, feature counting tools were tested (at least ten times) under fixed lighting and optical conditions, respectively.This is used to determine the tolerance level of the feature size.In a real-In life applications, the number of tests must increase due to the possibility of noise interference, such as lighting conditions that change over time, or other sources such as electrical signals.With well-Setting tolerance levels on minimum and/or maximum feature dimensions can achieve 100% efficiency during inspection.
Custom message