This paper describes the new digital image processing techniques for measuring pollutant diffusion around buildings. Calibrations of the image processor are performed for the linearity and stability of video signals, and for the shading of video images, using standard gray scale cards. Wind tunnel experiments are also conducted to investigate the averaging time of video image signals and the accuracy of these experiments. The vaporized oil is emitted into the wind tunnel flow from a stack. The smoke flow is visualized clearly by rapidly sweeping laser light sheet through the flow. The image pictures are sampled at intervals of 30 Hz. The following conclusions are confirmed. The averaging time for turbulent video image signals shows that 8 seconds is seen to be long enough for the running average to stabilize. For the accuracy of the experimental reproductivity, the difference error is obtained less than 18 %, which is considered to be accurate enough to measure the flow characteristics. Time-averaged tracer concentrations are compared with those for the video images. They are relatively similar with those for the video images. Consequently, the video image technology is demonstrated to be applicable to predict the pollutant diffusion around buildings, especially the temporal-spatial distribution, which is very much difficult to obtain from traditional point tracer measurements.