This technology is an image processing method that captures images from both visible light and infrared cameras, upsamples the infrared image to the size of the visible light image, and fuses it with detail edge data extracted from the visible light image using a high-pass filter to generate high-resolution infrared images.
While infrared cameras are useful for night vision by detecting thermal radiation, their resolution is significantly lower than that of visible light cameras. Even when using super-resolution techniques that utilize continuous images, there has been a persistent issue where fine details are not displayed clearly or are omitted entirely.
This technology extracts edges from visible light luminance images using a spatial filter matrix composed of negative peripheral coefficients and positive central coefficients, then adds them to infrared images enlarged via nearest-neighbor, bilinear, or bicubic matrices to enhance sharpness. It operates solely through filter calculations without requiring a separate learning process, making it simple to implement. It can be applied to automotive night vision systems equipped with dual visible/thermal cameras and industrial thermal inspection cameras, providing infrared images with clear object outlines even when using low-resolution thermal sensors.
N/A
