As a key indicator of vegetation, ground cover is an important parameter for Remote Sensing inversion. Researchers often employ visual methods to estimate ground cover, but these approaches may be prone to observer bias. An innovational method based on color features was presented to analyze digital photographs of vegetation to quantify percent ground cover as part of a study in Qinghai-Tibet Plateau. The k-means clustering algorithm was used to classify after processing the image pixel by pixel, and then calculate the ground cover. Compared the result with Supervised Classification, the difference between the two methods was fewer than 5%. Moreover, some approaches of improving accuracy of classification were discussed by analyzing the result image.