Development of an Application for LED Display Quality Diagnostics
The Customer is one of the world’s leaders in LED display manufacturing and distribution that contributed to the image of Times Square as an advertising Mecca. The factors of the company’s success include a strong monitoring functionality of display condition that evolves along with increasing needs of advertisers.
The Customer was looking to enhance real-time monitoring of the image quality on its LED displays with automated diagnostics via on-site USB/IP cameras. For this, the Customer turned to ScienceSoft, a provider of custom software development services, to get a user-friendly desktop application that would analyze information collected by cameras, detect and report problems with displaying advertisements.
ScienceSoft’s team delivered a standalone application running on Windows and Linux. The application uses computer vision algorithms to detect and classify problems with displaying images as captured by the cameras, reporting the results to the web server through HTTP requests in real time.
Once launched, the application analyzes each panel that makes up a LED display for errors, comparing its condition with other neighboring panels to detect if:
- a panel has frozen
- a panel has switched off completely
- a panel displays colors incorrectly because of under- or over-representation or the red, green, or blue channel
In each case, the analysis is based on the image processing algorithms that employ the OpenCV Library capabilities such as Gaussian blur, motion detection to identify frozen panels, mask application to detect the edges of a display and cut off irrelevant elements captured by the camera, as well as detection of colors that fail to correspond to the normal attributes.
The application checks for all the 3 error types at a time, and returns the pass/fail result based on a voting system to avoid false positives and false negatives in error reports. E.g., if a panel shows up as switched off but not frozen, it’s likely to be false positive and hence not reported.
ScienceSoft’s team delivered the application that could reinforce the Customer’s capabilities of monitoring the image quality of their LED displays in real time with accurate image processing algorithms.