Multistage median filtering method for smooth denoising of metallographic image

Abstract

The invention belongs to the field of image filtering, and specifically discloses a multistage median filtering method for smooth denoising of a metallographic image. The multistage median filtering method comprises the following steps: S1, taking a certain pixel as a central pixel on the metallographic image to obtain a plurality of filtering intervals, and acquiring a pixel gray value set in each filtering intervals; S2, respectively calculating a pixel gray value variance of each filtering interval, and taking the filtering interval with a minimum pixel gray value variance as an optimal filtering interval; and S3, performing median filtering on the central pixel by using the optimal filtering interval, and traversing each pixel in the metallographic image by using the central pixel so as to complete smooth denoising of the metallographic image. The invention provides a filtering method for smooth denoising of a plain carbon steel metallographic image. Thus, interference signals likenoise are removed; meanwhile, the boundary in the metallographic image is completely protected; and the accuracy of subsequent metallographic recognition work is improved.

Description

Technical field [0001] The invention belongs to the field of image filtering, and more specifically, relates to a multi-level median filtering method for smoothing and denoising metallographic images. Background technique [0002] In the inspection work of ordinary carbon steel castings, the analysis of the metallographic structure is mainly based on the recognition and judgment after the metallographic microscope image is collected. The metallographic image collection process is very susceptible to various electromagnetic pulses and noise interference, which will cause the identification The work caused great errors, so corresponding digital image processing methods are needed to smooth and denoise the extracted metallographic images of ordinary carbon steel to improve the accuracy of the subsequent identification of ordinary carbon steel metallography. [0003] The existing smoothing and denoising filtering methods, such as median filtering, mean filtering, etc., do not distingui...

AI Technical Summary

Problems solved by technology

[0002] In the detection of ordinary carbon steel castings, the analysis of the metallographic structure is mainly through the identification and judgment after the collection of metallographic microscope images, and the collection process of metallographic images is extremely susceptible to various electromagnetic pulses and noise interference, which will affect the identification The work caused great errors, so corresponding digital image processing methods are required to smooth and denoise the extracted metallographic images of common carbon steel, so as to improve the accuracy of subsequent metallographic identification of common carbon steel

[0003] Existing smoothing and denoising filtering methods, such as median filtering and mean filtering, do not distinguish between boundary signals and noise signals when processing pixel gray information in the neighborhood of each pixel. In addition to the noise in the metallographic image, the damage to the detailed information such as the metallographic boundary is also serious, which brings great errors to the subsequent metallographic identification work

Images