In-Store Video Analytics
IP surveillance video can be used for more than just security.
This is part two of a three-part series. The full version can be found in the Expert Column section of our website, Progressivegrocer.com.
One of the main arguments in favor of the move to an IP-based surveillance system is that IP (Internet protocol) technology can transform store operations in ways that analog systems never could. A strong example of this is the use of video analytics, which provide visual context around the information many grocers are already capturing, such as point-of-sale data.
At their core, video analytics are essentially software algorithms. Some create triggers based on "if/then" scenarios. For example, if the temperature in a refrigerator rises above a certain point, then a notice can be sent to a manager, along with the accompanying video. Giving a manager access to the video can help provide information that helps him or her determine the next steps to take.
Video analytics are also helpful at the checkout, where IP cameras can be linked to POS software. Louisville, Colo.-based Envysion, for example, has query-based software that enables video to be searched based on transaction type, store or employee. This includes no-sales transactions, voids, returns, and cash and credit sales. The software produces automated reports based on patterns and trends, which can then easily be matched to the corresponding video. Procedural errors — such as missed scans — are also captured, making investigations and overall management much more efficient.
Since IP video is really nothing more than data, it can be gathered, stored and analyzed just like any other electronic information. However, as with any other data-capture system, the accuracy of the output depends on the quality of the input. For that reason, the quality of the video is critical. The images need to be clear for the software to accurately analyze what's happening in each frame, such as distinguishing people from shadows for the purpose of people counting, classifying purchases by buyers' gender or age, or recognizing whether someone actually slipped and fell on store property.
When it comes to poor image quality, the most common culprit is lighting. While photographers can sometimes change their lighting conditions to produce better pictures, surveillance cameras are in fixed locations and need to produce clear video no matter how the lighting might change. Thankfully, IP camera technologies are addressing problematic lighting conditions by improving the "wide dynamic range" of IP cameras. This means that video is taken at various shutter speeds and then combined into a single video stream that makes both the dark and bright areas of the image more visible, which also improves the quality of the visual data that's fed into the analytics software, and makes the outcomes more accurate and reliable.
In addition to improving the dynamic range of IP cameras, reducing the "noise" in low-light or fast-moving video also enhances image quality and enables analytics software to be more accurate. In video transmission, noise refers to the level of static or blur in an image. Static often results from images that are poorly lit, while blur is a common problem with fast-moving objects against a still background. When using analytics, this noise often results in false positives, because software algorithms can interpret the pixel changes as motion. Today, digital noise-reduction technologies apply filters to video that effectively decrease static or blur. These help IP cameras and analytics software distinguish true motion from image noise.
Unfortunately, there are still no agreed-upon standards in the security industry that dictate minimum performance for dynamic range and noise reduction in IP cameras. Grocers, in turn, are still required to do thorough research and pilot tests to determine which cameras will produce the best image quality for their specific environments and lighting conditions. However, it's important to keep in mind that the deciding factor should be overall image quality, not the number of megapixels.
Mark Collett is GM of Sony Electronics' Security Systems Division in San Diego. He can be reached at mark.collett@am.sony.com.