Effect of different image processing techniques on the measurement accuracy of white-light interferometric fiber-optic temperature sensor

摘要

Images from a CCD camera may be degraded for various reasons. Several techniques for restoration of original interference fringes of a white-light interferometric (WLI) fiber-optic sensor are compared and analyzed theoretically and experimentally. These include vertical edge filter, background correction and FFT/Inverse FFT. Images are acquired from 2 cooling cycles between 50℃ and 20℃ with 1℃ step, respectively. Temperature control accuracy is ±0.02℃. All images are processed by these methods to identify the central position of the white-light interference fringes with a Gaussian profile. Error analysis result are given, showing that visually estimating image quality is not enough to provide higher measurement accuracy for some of these processing approaches. On the other hand, properly selecting FFT filters to attenuate and accentuate selected frequencies in the frequency domain will generate contrast enhanced interference fringes against a dark background, which will greatly improve accuracy and be less sensitive to image degradation.