In order to accurately acquire the tomato plants’ image information under the complex illumination in greenhouse, the method of correcting image color based on high dynamic range (HDR) imaging was researched, which was urgently needed for robotic production in greenhouse. Focused at the color distortion caused from sunlight’s continuous variation and background object’s radiation saltation, the color image’s brightness data was extracted from CIE XYZ color mode. And the camera’s response function was recovered, according to the brightness images with various exposure time of 0.01ms, 0.05ms,0.08ms and 0.10ms. As the HDR image data, the radiation intensity of the view field was estimated based on the response function, referring to the underexposed image’s brightness. The HDR brightness data was compressed into brightness image with grey value range of (0, 255), by the S-shaped mapping function, and then the brightness data was integrated into the underexposed image to reconstruct the image color data. Finally, the color correction method was verified by field test in greenhouse. As the result showed, the method was applicative for improving color quality of images, captured from different scenes under the various sunlight at different time. Specifically, the image’s entropy, standard deviation, and average gradient were averagely raised by 16.87%, 9.81% and 19.49%, respectively, after the original images captured with the serial exposure time were fused, and the color of image captured at different time could keep stable. The research result was supposed as the reference for acquiring object image information under the complex agricultural environment.