An analysis using infrared and visual images is made to measure the ice thickness of a cylindrical component. The proposed method is useful for ice detection and measurement on structures, even in harsh conditions and low light situations such as night. This type of analysis can fill a gap of knowledge related to ice measurement using both visual and thermal images. Thermal imaging shows differences in the emissivity and temperature of objects. This can help to detect objects and measure the amount of ice accumulated on the objects. Combining the information of visual and thermal images can compensate for their weak points and present better results. Combinations of the color-visual image (CVI), grayscale-visual image (GVI), color-infrared image (CII) and grayscale-infrared image (GII) are used to find the most accurate results. A binary image is acquired using the threshold method based on data collected from infrared and visual images. Using threshold levels removes irrelevant data that come from the background. Common ice pixels detected from both infrared and visual images are considered as the ice area. Thresholding methods cause unwanted gaps and strips in binary images. Morphological algorithms are used to remove these imperfections. The best results are obtained when one of the elements of the combinations is CII. The results of using CVI and GVI are almost the same. The experiments show that this method is reliable and its results are aligned with the real data.