For the problem of high-precision tracking of space area targets, a method of tracking area targets with high-precision was proposed. Sum of weighted absolute differences(SWAD) matching algorithm and sub-pixel fitting algorithm were used to extract precise position of the target. The template was updated during tracking. The conventional infinite impulse response filter approach used for template updating was improved. A variable coefficient template updating method was proposed. This method had small amount of computations, and creditability evaluation was not needed. The template updating coefficient was determined by gray values of current template and current best matching area. Images with different target brightness, and images after scale transformation for simulating target attitude change, were used to compare matching error of proposed method and conventional method. Results show that method proposed in this paper has better adaptability to changes of target attitude. Finally, collimator and simulation target were used to simulate far-field non-cooperative target. Indoor demonstration experiment was set up and the experiment shows that high-precision target tracking with image matching is feasible.