An Adaptive Threshold Method for Binarization of Fluorescence Microscopic Images

Abstract

The invention relates to an adaptive threshold value method for binarization of fluorescent microscopic images, which belongs to the microscope system image processing technology. Including: filter out low-frequency background, image background homogenization; calculate the normalized histogram of the image, obtain the normalized histogram curve; fit the power function curve; obtain the turning point of the curve by deriving the fitted curve; Perform binarization. Starting from the characteristics of low background of fluorescence microscopic images, the present invention provides an adaptive threshold selection method based on normalized histogram curve fitting, which is suitable for binarization processing of fluorescent microscopic images.

Description

Technical field [0001] The invention relates to the field of image processing of microscope systems, in particular to an adaptive threshold method for the binarization of fluorescence microscopy images. Background technique [0002] Fluorescence microscopy imaging technology is one of the important research methods in biomedicine, cell and molecular biology, microbiology, environmental science and other fields. [0003] In image processing, binarization is one of the main processing methods. The basic idea of ​​binarization for grayscale images is to determine a threshold, and then compare the grayscale value of each pixel with the threshold, and compare the results according to the comparison. Each pixel in the image is divided into two categories: target and background. For pictures with uniform illumination, the classic global threshold method can achieve better results. These methods include: threshold method based on the valley bottom between double peaks of the histogram, it...

AI Technical Summary

Problems solved by technology

For fluorescence microscopy images, the characteristics include: weak background light, the target has relatively strong fluorescence; background illumination is uneven, especially under high-magnification objective lenses, there is a significant brightness difference between the middle and surrounding backgrounds; when there are few targets ( Such as microbial microscopic imaging), there is no obvious peak in the histogram, and most pixels belong to low gray values; there may be multi-color fluorescent imaging signals at the same time, which interferes with the binary processing of the fluorescent target of the target color; Image quality varies

For fluorescence microscopy images, the commonly used global threshold method has practical difficulties in application. The bimodal method cannot be used for fluorescence images with no peaks in the histogram. The OSTU method and iterative method are relatively close to the background and target. SNR image processing failure

Images