A Study On Corrosion Detection For Q235 Steel In Seawater Based On Image Analysis

The development of computer technology and modern mathematical theory can provide a powerful technical and theoretical support for the detection and analysis of corrosion morphology on the material surface.This is of great significance to improve the automated assessment for corrosion characteristics of material surface;meanwhile it also has a broader application prospect.In this paper,for exploring a more convenient method to characterize the corrosion cumulativedamage of ships and marine structures,the characteristic values of the corrosion morphology image of Q235 steel were extracted and analyzed by employing the electrochemical measurement technique,corrosion image acquisition technique and also some digital image processing techniques such as grayscale transformation,image binarization and wavelet transform,etc.The main work and conclusions of this paper are as follows:The effects of stress on the corrosion behavior of Q235 steel in 3.5% NaCl solution were investigated by employing the electrochemical measurement and corrosion morphology image acquisition.The results showed that the main corrosion type of Q235 steel in 3.5% NaCl solution was local corrosion,and stress could contribute to its corrosion.Meanwhile,it was found that the linear polarization resistance measurement,electrochemical impedance spectrometrymeasurement and the electrochemical noise measurement could have a good consistency in characterizing the corrosion behavior of metal.In addition,the corrosion morphology images of Q235 steel under different stress levels were captured for the next image processing and analysis.The statistical characteristic parameters of corrosion morphology images,the areas and quantities of pitting and also the calculated percentage of energy value based on the sub-band images through wavelet decomposition were extracted and analyzed by employing the smooth filtering and image enhancement techniques combined with some digital image processing techniques including image binarization and wavelet transform,etc.The results showed that the method based on the statistical characteristic parameters of corrosion morphology images was limited in characterizing the corrosion behavior of metal.However,the method based on the area and quantity of pitting through the feature extraction of 8-nearest neighbor algorithm after the image binarization was effective,so was the method based on the percentage of energy value of the sub-band images after wavelet decomposition.Overall,the percentage of energy value based on the sub-band images through wavelet decomposition decreased with the increase of stress,indicating that the percentage of energy value based on the sub-band image was inversely proportional to the corrosion rate.This showed a good agreement with the conclusions of electrochemical tests.The fractional Brownian surfaces under various fractal dimensions used for simulating the corrosion morphology in actual engineering were generated by employing the midpoint displacement method and the mathematical model of fractional Brownian motion combined with the improvement of interpolation algorithm.Furthermore,the fractal dimension and area factor of the surface corrosion morphology of Q235 steel specimen were studied and analyzed.The results showed that the fractal dimension and area factor of the corrosion image could reflect the surface roughness of specimen,and both of them increased with the enhancement of the corrosion rate.Compared with the electrochemical test results,it was found that the fractal characteristic parameters extracted from the actual corrosion morphology image could be used to evaluate the corrosion behavior of metal,and also this method mentioned above showed a very good evaluated effect.