The subtle color distinction is the important function of electronic endoscope imaging diagnosis. However, after image acquisition, transmission and display, color distortions of intracorporeal organs or tissues occur inevitably, which are adverse to analyze image features accurately or to diagnose early pathological changes. A real-time color correction algorithm based on fourneighborhood and polynomial regression in YUV color space is proposed. Based on polynomial regression the color correction matrix is calculated in YUV color space according to the differences between standard values of color checker and measured values of that imaged by the endoscope. As the correction is only executed on U and V components in YUV color space, the defect that the color of corrected images in RGB color space will change along with luminance can be avoided, and then the stability of image color is improved. Owing to four-neighborhood processing, the signal-to-noise ratio of corrected images is enhanced and the processing speed of correction algorithm is accelerated. The average color difference is reduced from 0.3944 to 0.2850 by application of the proposed algorithm in high-definition electronic endoscope. A total of 17 frames per second can be achieved at the resolution of 1280 × 800, and the color characteristics of the image after processing match that of human visual system.