The authorities in China have started to take active steps in the fight against fake and tainted meat products, according to news website AsiaOne in May 2013. This occurred after a series of food scandals that have shaken public confidence.
In the same month, MSN News reported that a crime syndicate was busted along with 63 arrests being made. The crime ring was responsible for processing rat, fox and mink meats before selling them as mutton in Shanghai for US$1.6 million.
Not surprisingly, a survey conducted by market research firm Ipsos indicated that over 75 percent of respondents had concerns about the quality of food products, based on a report from China Economic Review in July. Because of this, demand for imports have increased due to better safety controls during packaging and processing, and the stringent product testing.
For domestic food manufacturers who want to continue to stay in the game, ensuring that their food processing and packaging facilities are well equipped to maintain food safety standards is important. The existing landscape is also an excellent opportunity for them to build their own reputations in a sea of scandals and fraud.
In a bottling facility that produces soft drinks, it is important to have quality control (QC) mechanisms that pick out production errors. An example of the latter would be bottles that have missing caps.
Clearly, delivering an uncapped bottle to retailers could end up with the contents spilling out and creating a mess along the way. Similarly, bottles whose caps are crooked or improperly screwed on could have the same undesired results.
Besides, bottles that have not been completely sealed could also allow pathogens or contaminants to enter, resulting in the likely spoilage of the product. Accidental consumption of the beverage would present a health threat to an unsuspecting consumer.
Another scenario that could arise is the capping of bottles with the wrong caps. This could be due to a number of reasons, such as negligence. For example, the operator on duty may have loaded on the incorrect caps, or the supplier of the caps could have delivered caps that bore the wrong brand or product name.
While this might not necessarily pose a serious health issue, the embarrassment from the mistake could negatively affect the reputation of the brand name and the manufacturer.
The problem would be further exacerbated in a situation where a contract manufacturer manufactures products for different brands. Delivering a batch of Brand A product sealed with Brand B caps could make a major laughingstock of the two brand names involved—especially given the speed and accessibility of online social media today.
This would naturally raise doubts in the consumer market over the credibility and integrity of the companies. It could also end in expensive lawsuits against the contract manufacturer and the potential loss of business and reputation.
Manufacturing and expiry dates are also typically printed directly on bottles. This is to help the supply chain and end users to identify products that have passed their useful shelf lives. In this manner, it can help prevent expired food and beverage products from being mistakenly consumed. And if this does happen and an adverse reaction occurs, it is easier to prove that the onus of responsibility does not lie with the manufacturer.
The manufacturer, however, has to ensure that the printed dates are correct, legibly printed and accurate. For example, an expiry date of February 30, 2105 could draw ridicule and ire (since there are only either 28 or 29 days in that month) from the public, putting the reputation of the manufacturer at stake.
Yet, all of the above-mentioned problems can be easily resolved with the help of machine vision. A particular manufacturer of beer initially depended on four human operators to perform QC operations on beer bottles travelling along the conveyor belt.
The operators had to manually inspect the bottles to ensure that all the bottles were capped and that the expiry and manufacturing dates had been printed. This method seemed to work initially when the facility’s output was still low. However, when orders increased and production speed was ramped up, problems began to surface that questioned the suitability of having human operators on this job.
At a production line speed of 50,000 bottles an hour, the human eye could hardly keep up. This meant that the inspection process was not thorough and not every bottle could be checked for defects. Also, while it was possible to check the presence of the dates, it was very difficult to ensure that the dates were correct.
After the vision system implementation however, every bottle is now thoroughly inspected, even at current production speeds. Any bottle that does not meet the grade is diverted to a rejection bin for collection by the operator at the end of the shift.
The system is also able to overcome another problem—verifying that the printed dates are correct. It achieves this by comparing the predefined dates that have been stored in the system.
With the previous manual checking process, operators were not able to record down any errors, leaving the management team unaware of production problems. Furthermore, problematic bottles that may have slipped through the earlier human QC system would have gone unreported, until and unless consumers submitted complaints.
Presentation Of Information
The machine vision system on the other hand, is able to generate reports on the number of errors that are detected during each shift, thereby providing managers with the necessary information to fine-tune and improve their manufacturing processes.
For example, if there had been a large number of errors with a particular batch, it could mean that there was a problem with a shipment of caps; or perhaps issues with the bottling equipment.
The machine vision system saves images of each bottle that has been checked and stores this data for future retrieval if required. These images are important in determining which bottles have the right caps on and which do not. In the event of any investigations or allegations of production error, this data can serve as important evidence.
Besides inspecting caps and dates, machine vision can also be applied to ensure that the labels on bottles are correct. Again, this serves as a safeguard against human error, whereby an operator may have loaded the wrong set of labels for a particular job.
Furthermore, the system also has the intelligence to ensure that the labels are not crooked or out of position. Machine vision can also determine if ‘double-pasting’ has occurred, ie: where a second label has mistakenly been pasted over the first. Although the surface of a bottle is rounded, the system is able to leverage on the three-dimensional reconstruction feature provided by OmniView technology—to generate one-dimensional images of the bottles.
This allows label checks to be made as if the product was flat. An example of a setup for this purpose, is the deployment of four cameras that can see the entire surface of a cylindrical bottle. This is regardless of the position that the bottle is in. The software processes the four images and displays them as a single image. At the same time, it compensates for any variations in position during the bottle’s movement along the line.
The technology eliminates the need for expensive investments to be made to mechanically rotate the parts for the inspection of such containers. In this manner, it is possible to reduce the complexity of the production line, lowering costs and increasing production rate.
The credibility of a brand name is an important consideration in the eyes of the consumer. The media has shown that reputations of companies that took many painstaking years to build, can be swiftly brought down by a single act of negligence.
In order to safeguard themselves, manufacturing facilities would do well to invest in automated QC systems that can prevent such events from occurring. This can help prevent negative publicity, ensure product quality, while building up the reputation of a company.
Light For Sight
In vision applications, the wavelength of the light that is used may be an important variable in the inspection process. When light of a certain colour is used for illumination, contrasting colours can be differentiated more easily. For example, this method can be used to enhance the date codes found on jar lids.
Infrared light can be applied to highlight bruises on produce, while ultraviolet (UV) light will cause ink or glue to fluoresce. Filters need to be used on the camera if infrared or UV lights are used. This allows the fluorescence of the object to be viewed, but not the light itself.
Collimated light is created when light rays are aligned parallel, and is suitable for applications that require sharp images. Diffused light is produced by directing collimated light through frosted glass. This results in a softer, more even illumination which avoids glare or shadows, but with the downside of lowering the intensity of the light.
Another consideration when inspecting parts is the speed of motion of the part (eg: when bottles are moving along a conveyor belt) as they are photographed, as well as the exposure time of the camera.
The choice of light source therefore depends on a number of production variables. By defining the goals and focusing on what is important for the inspection, the application can be simplified. Once these goals have been understood, choosing the light source is the next step, depending on whether the requirement is to highlight features/defects or to cancel out the unwanted elements of the image.