Researches have pointed out that the colors have the function of conveying messages and is even easier than words to be memorized. The color image of a city, which means people connect with color through knowledge acquired and their life experiences. It includes landscapes, buildings, food, the city culture, local specialties and so on. Thus, building up the distinctive style of a city is an important part of the city image and its related applications. Nowadays, people mostly use three different applications in order to set up and correct the city color tickets. First, field research; second, residents participate in the comprehensive community development; third, particular projects. In this paper, we use data mining to analyze the connecting between people and the city color image. At the beginning, we select ten cities of Taiwan, which were elected by Taiwan Tourism Bureau as our research object. Secondly, we use Word to vector and Google searching engine to find the relevance between the city and adjectives. For the third step, the highest connection of adjective will be the keyword of Google searching engine to collect the pictures by doing cross-comparison of the results. Last but not least, we capture ten colors from city pictures and result in city color combinations. According to the result of this paper, although it needs to adjust in accordance with local culture, we still can find the regional pictures that correspond with the characteristics and also can obtain the city color combinations, which can be used as a reference of harmonious colors.