GISdevelopment.net ---> AARS ---> ACRS 1992 ---> Digital Image Processing An Attempt to Observe Paths of Particle of Wind Flowing over Buildings Utilizing Simplified Aerial Photogrammetry Toshio Koizumi, Hiroto Hagura Chiba Institute of Technology 2-17-1 Tsudanuma, Narashino-shi Chiba, Japan Introudction It is important in the study of wind to three-dimensionally determine on site the flow of wind near the ground surface. The authers developed a system to measure the data of wind flow in both the horizontal and vertical directions utilizing aerial photographs. This system uses two cigar-shaped kite balloons to which is mounted a 35 mm still camera, which stereophotographs a third balloon release as a tracer, thereby three-dimensionally analyzing the third balloon's path of particle. This time, we utilized a system in observing the paths of particle of wind flowing over a building, and confirmed the effectiveness of the system. Observation System This system was developed for the purpose of observing paths of particle of wind flowing. Fig. 1 shows outline of this system. Fig 2. Devices and Analysing System for Aerial Photogrammetry Utilizing kite Balloon This system I composed of following elements as shown in Fig. 2. 1) A kite balloon born camera system. - Balloon ; 7 m3 , 4.25m3 , helium gus - String ; 1,000 meter - Still camera ; 35mm, wide angle, f=28mm (Olympus 0M-1, Canon T70) - Monitor camera ; Pentax PC-K1200, Sony CCD camera - Radio control ; Sutter, monitor and angle control Horizontal 360 degree Vertical ± 90 degree - Time lag of two camera shutters is less than 0.04 seconds. - Total payload ; 5.0 kg, 3.0 kg. 2) Analytical system - Coordinate digitizer ; Resolution 0.025 mm ( Mutoh BL ) - Personal computer ; NEC 9801VX Fig. 3. shows the signal sending and receiving systems. Observation of paths of particle of wind flowing over buildings. The observations were made with buildings in campus of Chiba Institute of Technology. Fig. 4 shows the site of experiment. Fig. 5 shows drift route of balloon. Fig. 6 shows release of balloon. Observation of wind direction and velocity when photographed were made with aerovanes mounted on a 200-ft tower. Table. 1 Values of Wind Direction and Velocity When Photographed Test Wind direction Wind velocity (m/s) 1 SW 2.1 2 ENE 3.7 3 NNE (Before 10 minute) 3.3 (Before 10 minute) In test 1, the balloons were released from the hill as showing Fig. 7 and took thirty frame pictures. In test 2, the balloons were released from the area between building No. 2 and No. 3 as showing Fig. 8 and took thirty-three frame pictures. In test 3, the balloons were released from the roof of building N. 5 as showing Fig. 9 and took twenty-seven frame pictures. These pictures were printed 16.5 cm x 24.5 cm size. Three-dimensional coordinates of the balloons released as a tracer were computed using coordinate digitizer and personal computer. Results of the test shows from Fig. 7 to Fig. 9. Conclusion It was experimentally confirmed that, using this system the paths of particle in both the horizontal and the vertical direction could be observed on site simultaneously. It was confirmed that the system was suitable for fixed point observation at any altitude and that unskilled personnel could utilize the system simply and safely. Consequently, the photographing of urban areas, etc. can be made, and the system is conceivably suitable for three dimensionally observing winds including those flowing over buildings. From the charts of the paths of particles obtained, it was found that it is possible to confirm the behaviors of winds the velocity of which increases at the windward eaves of a building as the data observed on site. mean standard deviation of the second image. Calculate the different image obtained from two images. Create 2-dimensional histogram which composes of the preferred removal cloud and shadow image (Fig. 1 (c) ) in first axis and the different image of Fig. 1 (c) and Fig. 1 (b) in the second axis. Find the first group thresholds T11, T12 , T13 and T14 , for detecting the cloud pixels, by interactive checking with monitor. Find the second group thresholds T21 , T22 , T23 and T24 , for detecting the cloud shadow pixels, by interactive checking with monitor. Create a cloud and its shadow map. Fig. 1 Two visible satellite image with different cloud distribution are show in (a) and (b), while (c) is the resulting image from brightness matching of (a) to (b). Replace the cloud pixel and its shadow by the brightness of the corresponding pixels from the second registered image. Result Two registered image from MOS I satellite have been used as shown in Fig 1 (a) and (b). while Fig 1 (c) is obtained from brightness matching of Fig. 1 (a) to (b). 2-dimensional histogram, which compose from the Fig. 1(c) and the different between Fig. 1(c) and (b), is shown in Fig. 2. To detect the cloud pixels, the first set of threshold have been used T11 = 93 , T12 = 255 T13 = 10 and T14 = 210 which give the green corresponding area in the histogram. Which the cloud shadow pixels can be detected by using the second set of threshold for T21 = 67, T23 = -81 and T24 = -5 as shown in the Fig. 2. which given the red corresponding are in the histogram. The resulting map of cloud and its shadow distribution is show in Fig. 3. The final result is cloud and its shadow removal is shown in Fig. 4. Fig. 2. 2-dimentional histogram formed by the first image and the difference image. The green are is corresponded to cloud area while the red area is cloud shadow. Conclusion The method of cloud and its shadow removal has presented. This method can be accomplished if the cloud and its shadow have different distribution in two images. This method can removed not only the first image but also the second image the result demonstrate the feasibility of combining threshold detection in the mosaicking to remove cloud and shadow. This method can be extended to create surface type maps such as sea, land and cloud etc. Reference "Manual of Remote Sensing", Vol. 1, The American Society of Photogrammetry. .J.A. Parikh and A. Rosenfeld, "Automatic segmentation and classification of infrared meteorological satellite data", IEEE Trans. Syst. Man Cybernet., Vol. 8, pp. 736, 1978. R.G. Crane and M.R. Anderson, "Satellite discrimination of snow/cloud surface", Int. J. Remote Sensing, Vol. 5, pp. 213, 1984. A.M. Seddon and G.E. Hunt, "Segmentation of clouds using cluster analysis", Int. J. Remote Sensing Vol. 6, pp. 717, 1985. L. Zhirong and M.J. McDonnell, "Technical note : Using dual thresholds for cloud and mosaicking", Int., J. Remote Sensing, Vol. 7, No. 10, pp. 1349-1358, 1986. Z.K. Liu and B.R. Hunt, "Note : A New Approach to Removing Cloud Cover from Satellite Imagery", Computer Vision, Graphics, and Image Processing Vol. 25, pp. 252-256, 1984. P. Zamperoni, "Contour tracing of grey-scale images based on 2-dimensional histogram", Pattern Recognition, Vol. 15, No. 3, pp. 161-165, 1982.