Inventor(s)

Tony ChuFollow

Abstract

Smartglasses and other wearable devices can automatically detect the presence of text in the environment via images captured by a high-resolution camera that is turned on periodically. Such an approach is computationally intensive and consumes large amounts of power. This disclosure describes techniques for low-power text detection on an AR/XR/MR capable device by augmenting the device to perform text detection via a low-resolution, ultra-low-power auxiliary camera with a lens profile that captures greater detail in the center. Distorted low-resolution images from the auxiliary can be de-wrapped and/or cropped as needed and analyzed l using machine learning models to detect the presence of text. The smaller image size provides reductions in the amount of data captured, transferred, and analyzed, resulting in a corresponding reduction in power consumption. A higher-resolution image can be obtained via the main high-resolution camera as needed for subsequent actions, such as extracting the text via OCR.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS