There are usually a series of processes in the development program of machine vision applications, including: planning, design, build, integrate and verify.
The first step in any solution is to determine the requirements of each inspection and determine if feasible, so plan is the primary consideration. Several factors need to be considered, such as: what is being inspected, to limit the number of required inspections of speed, mechanical design, performance requirements, time and cost.
After the planning phase is past the design stage, to ensure that it meets the requirements, the need for prototyping. If you think that engineers can achieve visual inspection may not be needed for prototyping. According to inspection requirements, create and test the visual design for each of the initial vision applications. There may be several visual stations, all stations are visually different optical settings. In the design of visual workstations, each of cameras, lenses and lights are several factors to consider. Camera specifications, the number of cameras, camera type, monochrome or color, resolution, frame rate, communication protocol, exposure speed, lens specifications, lens type, focal length, working distance, interval, field of vision, filtered, aperture, lamp size, lamp the number, type of light, size, color, and distance part, strength.
Once you have identified a number of factors, we established visual design prototypes to capture the image of the sample portion. The software must be detected by having an appropriate key features in the image contrast. This procedure involves adjustment of the optical device.
The image of an array of pixels, typically defined by the resolution. Machine vision software using these pixels and a predetermined algorithm to define certain features of the image member. It follows the same concept of facial recognition software. The software searches for certain pixels are arranged in the image to identify facial features. For example, the peripheral edge of the eye will be darker pixels, the center of the eye will be darker pixels. The software uses the algorithm analysis feature, and facial recognition. Similarly, the machine vision software follows the same concept. These algorithms use machine vision software developed by visual engineer.
There may be some mechanical limitations, which would require additional optical devices, such as mirrors. There may also need to change the design of mechanical constraints. For example, since the automated tools around the optical device, the working distance may be required between 300mm and 320mm. There may need to replace the lens, while increasing the extension tube.
If the image does not meet the requirements, troubleshooting. Change various parameters to allow a more suitable image. You may need to be replaced or lighting may need to add the polarizer. Prototype design allows engineers to visually verify the visual design and create the visual design.
Conventional machine vision cameras usually require visual controller. If you are using smart cameras, you do not need an external visual controller. The controller is a dedicated unit for visual communication with the optical device (e.g. camera and light). Vision controller with the capability to meet the inspection requirements. This means it must be able to check for a communication processing apparatus, and other I / O protocol is required.
At this stage, it should be clear whether you can meet the requirements of visual inspection. The software should be able to check the required detection process. To verify that the software is able to detect edges, use that will be used during the inspection some tools (ie, edge positioning tool) to ensure that the software can detect certain edge. After this attempt to change the design could increase the cost of the solution. Ensure confidence in the visual design at this stage.
Development of
machine vision software can use a variety of tools to help feature analysis part. Before building codes first thing to do is to use the example images to plan ways to build code. It will also depend on what kind of machine vision software. Using the designed settings, good and bad gripping several partial images, for use at this stage.
Different software may have a different set of tools. Some software may not be able to keep up with automated cycle time. If you need to quickly check the cycle time, you must complete a comprehensive examination at that time. It's all back to the inspection requirements, to decide which software to use.
The inspection requirements, when building codes need to consider several factors. Some examples include: Each member of each camera capturing a plurality of images requires it? Need to check the function of what? Strobe lights need more frequent? Check which tools need to use? How to calibrate? Whether the software is communicating? How to ambient lighting play a role? If there is a light inspection, this lamp when to trigger? Does it still exist throughout the inspection process? It will flash it? If you need to use the camera to trigger the lights, you need to use software to solve. If it involves strobe unit, the unit strobe How often?
We need to perform a series of software tools for inspection. Accordingly also need to place tools to ensure that the software checks only need to check the function. Using these tools need to have a certain logic. For example, if the software defect is detected, then the fault, which output an error component generated code and display the results.
Any need to send this information to an external controller for use with automation. If the part is a pass or fail, it will allow the machine to know how to handle parts. Through these communications, but also I / O functions to communicate with other automation.
By communication, but also needs to have a user interface (UI) for the operator readily understood. This means ensuring that all results are easy to see, the operator can easily use any other function. The error code should be easy to understand for the operator.
For accurate measurements, a calibration step is required, so that the software features in the image can be measured. One method is to use a calibration grid. Acquired from the visual image of the calibration grid station to allow calibration check. Most machine vision software has such a tool. The tool can be used to select the calibration grid, it will ask the parameters of the grid.
Integration
integration phase is all optical devices integrated with other automation places. It's everything together into a whole place.
The first thing to do in integration phase is to ensure that all hardware are configured and installed on the automation components. Design specifications adjusted according to each stage of the design vision station. To ensure that all communications between all devices are normal.
Using the previously developed calibration step, calibration of the optical device. If the calibration procedure is completed, use the calibration target camera view in this step and run to capture the image. The software calibration to the calibration measurement. Ensure that the same measurement value and the actually measured value of the software part. If a lamp is used to check, adjust the light intensity to meet the desired luminance.
By triggering multiple images of optical test equipment. If you read a good image, fully automated optical test equipment. Good and bad test components to ensure that the part rejected. Ensure that all automation and time synchronized with the optical device, to ensure effective inspection.
If the check at this stage does not meet the requirements, troubleshooting. According to the problem, there may be content code needs to be changed. At this time, change the code is a simple fix. If you need to replace the equipment or any other significant changes, it may increase the time and cost of the solution.
Verify
at this stage, we will conduct a number of tests to ensure that the vision station to meet the requirements outlined in the beginning of the project. The test is based on the inspection requirements created. Each time a test is visual inspection should be carried out to test repeatability. A good example of this test is to gage repeatability and reproducibility tests. Changes and variations of the test member to check the measured value.
If the requirements are not met, please troubleshoot again. To try to fix by checking to make minor changes. It will take less time and resources. In this phase of the project, significant changes will take more time and resources. Once all tests passed and meet all the requirements, the solution is complete.