An algorithm was developed for identifying and tracking natural vibration modes. Its main feature is the using of image operations. This enables the pattern recognition of natural vibration modes and the automatic surveying of parameters in vibration design even when the change of order in which natural vibration modes appear occurs. The developed algorithm consists of the following steps. First, the absolute value of each component of a vector corresponding to a natural vibration mode is represented as a gray-scale image of each component by a computer post processor. Each of the gray-scale images is then binarized with white and black colors. The numbers of the groups consisting of pixels with the same color are counted in each binary image, and the numbers are used for distinguishing natural vibration modes that should be tracked from the others. A structural optimization system was developed that contains a software routine using the developed algorithm. The reliability of the developed algorithm was examined through some calculation examples.