OCT (Optical Coherence Tomography) image speckle noise reducing algorithm based on adaptive bilateral filtering

Abstract

The invention discloses an OCT (Optical Coherence Tomography) image speckle noise reducing algorithm based on adaptive bilateral filtering. The algorithm comprises the following steps of: establishing a speckle noise model of an original OCT image; establishing a space function by taking the speckle noise model of the original OCT image as a variable according to the Rayleigh criterion; and analyzing the characteristics of the space function and deducing a formula for adaptive correction of a filtering weight coefficient W[k, i] by the space function. The OCT image speckle noise reducing algorithm based on adaptive bilateral filtering disclosed by the invention can reduce the speckle noise of OCT image, reduce the mean square error of image and increase the peak signal-to-noise ratio; and meanwhile, the algorithm remarkably maintains the edge information of the image, increases the edge contrast and obtains clearer image edge details.

Description

technical field [0001] The invention relates to an OCT image speckle noise reduction algorithm based on adaptive bilateral filtering, and belongs to the field of OCT image processing. Background technique [0002] Optical coherence tomography (Optical Coherence Tomography, OCT) is a broadband optical scanning tomography technology developed in recent years, which uses the low coherence of broadband light sources to achieve high-resolution, non-invasive optical tomography. The exploration of OCT technology began in the late 1980s. In 1988, A.F.Fercher et al. used low-coherence interferometers for intraocular measurements. In 1991, D.Huang and others proposed the concept of OCT, and obtained retinal images with this technology. Subsequently, research groups around the world have conducted many fruitful researches on OCT technology in terms of improving image resolution and signal-to-noise ratio. At present, the resolution of OCT technology can generally reach more than ten m...

AI Technical Summary

Problems solved by technology

For the bilateral filtering algorithm, choose an appropriate σ r and σ d The noise can be reduced to a relatively small range through one-step calculation, but since this algorithm is a local denoising algorithm, for OCT images, σ r The choice of σ is very sensitive: if fewer σ r , then there will be more singularities whose gray values ​​are significantly different from their neighbors in the denoised result image; and choosing a larger one will make the denoised image excessively smooth

If you want to get the ideal denoising effect, you need to adjust σ r (t) and σ d (t) is adjusted accordingly over time, but an appropriate σ r (t) and σ d (t) The function construction is more difficult, and even if you choose better denoising parameters, some faintly visible noise patches will appear in the denoising results

It is precisely because of these problems that the conventional bilateral filtering algorithm is unsatisfactory in the effect of OCT image speckle reduction.

Images