Rosa Damascena or Mohammadi flower in Persian, is a very drought-resistant plant that can withstand various climatic conditions. Considering the water crisis that has intensified in recent years and also the Agricultural Ministry of Jihad's policy for expanding the low water consumption crops, it is advisable to use drought-resistant plants such as Rosa Damascena for cultivation in areas with low rich soil and water. Flower harvesting is the most sensitive and important step, since after opening, the flower has a little durability on the branch, and with delay in harvest, its color will change to white and its petals will fade within 24 hours. Moreover, the faster the flower is harvested, the more quality and quantity of essence will be obtained. Due to the insufficient workman in the harvesting season, it is highly necessary to mechanize the harvesting of Rosa Damascena . This research focused on design an intelligent system to detect the flowers using image processing technique. In this system, two perpendicular digital cameras, a camera holder and a laptop were used. After capturing the top and bottom images of the flower using the cameras, the images were transferred to the computer and the flower was detected from background by running an image processing program written in MATLAB software. However, another method was proposed in this study, because of the some errors to accurately determine the location of the flower, as well as the relatively long processing time. In the new method, the video was captured using a mobile phone mounted on a camera holder. Subsequently, through the IP Webcam software, images were transferred online from the mobile to the computer. By specifying the hue and saturation values appropriate for flower color and also the ambient light, the flowers were then identified online by using the program written in MATLAB. To evaluate the system, three types of flowers were selected as three different attributes in terms of size and color. The factorial design with two treatments of robot's speed and robot's distance from flower, which consisted of two and three levels, respectively, were used in three replications. The result showed the ability of system for detecting 83% of flowers at distances of 100 and 120 cm and speed of 0.15 m s -1 . Increasing the speed reduces the ability of recognizing the flower for the system, but the increasing the distance to a maximum of 120 cm would not be a problem in diagnosis. Key Words: Rosa damascene , image processing, machine vision, color systems