This technology is a remote meter reading method and device that periodically captures images of analog meters, isolates the reading display as a region of interest, segments it into individual digit images, and uses lightweight machine learning to recognize the digits, transmitting only the numerical data wirelessly instead of the images.
Remote reading for water, electricity, and gas has traditionally required replacing analog meters with smart meters, creating a significant cost burden by forcing the replacement of functional equipment. While analog meters can be captured via camera, low-speed wireless devices face transmission capacity limits that prevent the sending of image files.
This technology automatically segments the black digit window by converting images to grayscale, quantizing them, and tracing boundaries from the vertical centerline. By using contour-based cropping and resizing to lighten the digit images, it enables recognition on low-spec devices. Since only the recognized digits are transmitted, it supports long-term, low-power operation. It can be applied to services that digitize existing analog meters without replacement, allowing for accurate data collection even in hard-to-reach locations.
N/A
