Case study: Software checks dishwasher integrity
Hobart Corp. (Troy, OH, USA; www.hobartcorp.com), a manufacturer of food equipment and systems, supplies ware-washing equipment to commercial organizations and consumers.
Hobart Corp. (Troy, OH, USA; www.hobartcorp.com), a manufacturer of food equipment and systems, supplies ware-washing equipment to commercial organizations and consumers. "Left to their own devices, bacteria will double every 20 minutes. In an hour, they will increase tenfold. In four hours, a single bacterium will multiply into 4000," says Chuck Warner of Hobart. "Because of this, our ware-washing products are designed to clean and sanitize plates and cooking utensils using features such as a 180° final rinse that is designed to kill germs and bacteria." To check the integrity of such machines and possibly improve future designs, Warner needed a way to quantitatively define the cleanliness of products placed in the ware-washing equipment. Warner turned to Futaba Corporation of America (Huntsville, AL, USA; www.futaba.com), which supplied Hobart with a CCD-based camera, PCI-based frame grabber, and the company's PC-based RoboFlo software.
"To measure the efficiency of the ware-washing equipment, a test plate is first stained with a circular fluorescent dot pattern," explains Warner. "Before the plate is washed, images of the circular dot pattern are imaged and their size measured. To do this, a test stand was built using fluorescent lighting to illuminate the plate. Images captured from above the test plate are then digitized and processed."
Jim Mandras, technical and application support specialist at Futaba, used RoboFlo to build the application program for Hobart. "RoboFlo's Windows-based architecture and visual top-down process flow coupled with the software's built-in machine-vision algorithms allowed a program to be developed that performs image calibration, measurement, and analysis within a few days."
RoboFlo features a number of calibration functions that correct a vision system from the distorting effects of camera angle and lens aberration. "The program first calibrates the vision setup prior to digitizing the images of the test plate," Mandras says. "This makes it possible for companies such as Hobart to use standard vision hardware, such as a XC-ES50 1/2-in. CCD camera from Sony Electronics (Park Ridge, NJ, USA; www.sony.com) and Meteor-II frame-grabber card from Matrox (Dorval, QC, Canada; www.matrox/imaging), lowering costs further." Without calibration, the rows and columns of dots in the calibration pattern appear somewhat distorted in the digitized image as the grid is visually scanned from side to side and vertically. Before running the program, the user specifies the number of rows and number of columns in the calibration grid along with the linear spacing between the centers of consecutive dots.
After completing the calibration procedure, RoboFlo prompts the user for a plate identifier, such as Plate2043. RoboFlo provides a live image and instructs the user to position a plate of dots under the camera for testing. Here, a set of 10-mm-diameter solid disks arranged in a 7 × 7 pattern simulates the dots before washing. After the plate has been placed under the camera, a BlobFinder vision tool within the program scans the digitized image and marks all the dot blobs found.
Because the system must analyze the dot pattern before and after washing, Mandras uses a similar grid of 7-mm-diameter disks to simulate the "post-wash" dots. The program features a system interface that presents a menu that drives the application. When the user clicks on the Pre-Wash Analysis button, for example, RoboFlo performs a blob analysis of the dots before the plate is washed.
RoboFlo highlights all buttons at occurring cycles in green. In this way, the user can tell at a glance where execution is occurring. Should an error occur during a test run of a newly developed program, RoboFlo highlights the buttons for the cycle where the error occurred in red, so the user can locate and correct the problem. This reduces program-development time, as well as factory downtime.
Moreover, RoboFlo can control several tasks at once, increasing productivity. As Mandras points out, "You can set up a process that involves several tasks: vision cameras inspect products, a batch of previously inspected products is being labeled, while a new batch of products to be tested is being loaded onto a pallet. The software lets you synchronize all these tasks so that they are performed in a concurrent manner to speed up the overall process."
When execution reaches an AcquireImage button, RoboFlo retrieves an image from the camera or from computer disk. When conducting a blob search, RoboFlo can search for light blobs against a dark background or vice-versa.
RoboFlo then stores the results of the blob search (all blob data) into another variable that can be used as input for other useful blob-analysis functions. In addition, the software can sort blobs in order of their x or y coordinates or their areas. Data generated are stored in a comma-separated value file that contains values in a table as a series of ASCII text lines organized so that a comma from the next column's value separates each column value; each row starts a new line. Because of this, such ASCII-generated data can be read into a database application such as Microsoft Excel. The PC can then access and process these data files.
After pre- and post-wash data are collected, the information is stored in database format for later analysis. The program computes the areas of all dots, as well as the change in area for each dot. The resulting values are in the same unit as used for the dot spacing mentioned earlier.
Hobart had investigated other vision systems and found that RoboFlo had the smallest learning curve. "Previously, Hobart staff was measuring the dots using a manual procedure," Mandras says. "RoboFlo enabled the company to obtain data and perform analyses faster and more accurately." A statistical analysis of these data provides a means of measuring the effectiveness of the ware-washing process. According to Warner, the data can, in turn, provide designers with a means to more effectively develop future ware-washing equipment. — Andrew Wilson