Recently, drones have been used for many purposes. The investigation of tree growth in a forest is also one of the expectations for drone applications. The aim of this paper is that a drone recognizes a color marker which is set to a tree from an image, and a drone approaches to that tree, and then return to the goal zone. To solve the task, we propose to apply the algorithm of Particle Filter for the color marker recognition. The difficulty of applying image processing at a forest is that degree of luminance changes drastically, due to incident light through the trees. To overcome this difficulty, we attempt the method to update the tracking color by a time period. From the results of experiments, our proposal has been able to deal with changes in luminance intensity, and then a drone with our proposal has solved a part of the task for competition.