New Method for Eliminating Background Noise in Characteristic Spectral Imaging

Abstract

Abstract We proposed a method to eliminate the background noise, which is based on subtraction of two different central wavelength filters according to the characteristic spectral imaging. In the experiment, the sodium yellow light with 589 nm wavelength is selected as the characteristic wavelength of spectral imaging experiment, and the visible light industrial camera was used to collect diffuse reflective wall images. Firstly, two narrow-band filters with a center wavelength of 589 nm and 579 nm are placed in front of the camera separately, and two images of the signal light containing the narrowband characteristic wavelength are acquired. Then, the surf algorithm is used to find the feature points of the two images, and the two images are matched, geometrically transformed and cropped to obtain images of the same size and the same angle. After the histogram matching, the background gray values in the two images are close, and the two images are subtracted, which can obtain an image with removed background noise. The experimental results show that this method can provide a simple, high-speed method to effectively eliminate the background noise of the characteristic spectrum.