In order to investigate and predict the corrosion behaviors of steel wires in stayed cables under cycling loading and eroded environments, several corrosion images of different corrosion degrees and their corresponding weight losses, nominal tensile strength as well as percentage elongation after fracture are obtained through salt spray accelerated test.The adaptability of HSV and RGB color patterns in characterizing the overall color and local edge contours of wire-steel corrosion images is discussed by using wavelet decomposition technique, and correlational analyses of mean values and percentages of L1 norms of corrosion image wavelet coefficients associated with weight losses are conducted.A regression model is suggested to predict the weight loss of wire-steels.It is found that local pitting corrosionon steel wires can be obviously observed under combined effect of cycling loading and eroded environment, and the nominal tensile strength and percentages elongation after fracture of steel-wires decrease with their weight loss increasing. Wavelet analysis results of steel-wire corrosion images show that the change of corrosion morphology mainly include the changes of its color saturation and brightness, saturation component in the HSV color mode can better describe the change of the overall color of the corroded image, and the blue component in the RGB color mode can better reflect the change of the local topography of the corroded image.There is a good linear correlation between the corrosion image of wire-steel in cables and its weight loss as well as nominal tensile strength, which can be used to predict the weight loss and nominal tensile strength of corrosion wire-steels.