The non-rigidity of the chromosome enables it to have unpredictable shape and size at the time of imaging. Directly performing chromosome classification and pairing would cause large errors and low efficiency, while the operation of the straightened chromosome could not only improve the accuracy and efficiency of classification and pairing, but also be conducive to the diagnosis of chromosomal abnormalities. In order to effectively solve the problem of curved chromosome straightening, an effective chromosome straightening algorithm was proposed, which could be used to straighten chromosomes with different degrees of curvature. The algorithm mainly used the refinement method to extract the central axis of the curved chromosome and correct it. By recording the adjacent 8 pixel points of the point on the central axis, the overlapping area of the vertical line of the fitted line and the original image was obtained, and then the central axis was followed. The pixel was stretched, and the pixels of the overlapping area were moved to the corrected chromosome picture according to the relative spatial position of the pixels in the overlapping area during the stretching process. By comparing the renderings of the three straightening algorithms, the method proved better realization of the straightening of curved chromosomes.