Vol. 52 No. 1 257 268 (Jan. 2011) 1 2, 1 1 measurement. In this paper, a dynamic road map making system is proposed. The proposition system uses probe-cars which has an in-vehicle camera and a GPS receiver. On the probe-car, the system detects global position of roadside edge and road side hue information from the images taken by the camera. On the server side, the system integrates probed road shapes using DP matching based on hue information. The experimental results are shown that the proposition can generate accurate road map from plural road shapes detected by probe-cars. GPS GPS DP A Road Map Making Probe System by Integration of Road Shapes with Roadside Hue Information Hitoshi Yamauchi, 1 Akira Tomono 2, 1 and Akihiro Kanagawa 1 Recently, many car navigation systems are used. And the road map are used in the car navigation systems as a main information. However, the map data may not be refreshed and the contents is remained in old information. This is caused by huge cost for refreshing. For this problem, some researches are proposed. For example, a specific measurement car is employed for high accurate 1. PND Portable Navigation Device 3 1) Google Maps 2) 3),4) 1 Faculty of Computer Science and Systems Engineering, Okayama Prefectural University 2 Graduate School of Systems Engineering, Okayama Prefectural University 1 Presently with System Enterprise Co., Ltd. 257 c 2011 Information Processing Society of Japan
258 5) GPS 1 GPS 6) GPS DP 7) 8) 10) GPS GPS 2 3 4 5 2. 2.1 3 1) GPS Global Positioning System
259 Google Maps 2) GPS 3),4) GPS 11) GPS 6) 2.2 GPS 8) GPS 1 Fig. 1 Tracking line and road edge point detection. 9),10) 2 1 GPS GPS 2 2
260 3. 9) 10) ICP 3.1 2 GPS Fig. 2 Position estimation of pictures using GPS interpolation. ICP Interactive Closest Point 12) ICP ICP 2.2 GPS 3 3.2 GPS 9) 10)
261 Fig. 5 5 An example of getting road side information. 3 Fig. 3 The model of proposition for dynamic map making system. 4 Fig. 4 Road side region and hue sampling area. 4 m n h 0 h<360 i 1 i n H i { H i = h 360 } 360 (i 1) h< n n i (1) 3.3 RANSAC m 5 Hough
262 RANSAC RANdom SAmple Consensus 13) RANSAC RANSAC 3.4 3 GPS GPS DP Dynamic Programming 14) DP 2 2 DP 2 6 2 DP 6 DP Fig. 6 DP matching. w A A B A A = wa +(1 w)b (2) w 0.5 1 1 (2) w
263 H A,i = wh A,i +(1 w)h B,i (3) 4. 1 Table 1 Experimental data. data1 2007 1 19 09:30 2,923 data2 2007 1 25 09:30 2,946 data3 2007 1 31 09:15 3,152 datar 2007 1 18 14:30 2,888 1 GPS PointGrey Research Flea 1/3 CCD 640 480 [pixel] IEEE1394 13FM22IR 2.2 [mm] 15 [fps] GPS 48 3.3 [m] CEP Intel Core2 Duo 3.0 GHz 2GB RAM PC GPS [deg] [m] 5 36 00 134 20 60 30 m =43 n =36 15 30 50 100 [m] RANSAC 43 7.0 [m] w =0.7 4.1 1 1.7 [km] data1 data3 3 ICP 30 40 [km/h] datar GPS GPS GPS data1 data3 7 7 GPS 8 Google Maps datar GPS GPS
264 7 data1 data3 9 data1 data3 Fig. 7 Road shapes of data1, data2 and data3. Fig. 9 Integration result of data1, data2 and data3. Fig. 8 8 data1 data3 GPS 1 GPS measurement result of data1, data2 and data3 (magnified, no.1). 2.5 3.0 [m] datar 7 GPS data1 data2 1 data3 2 9 10
265 Fig. 10 10 data1 data3 Integration result of data1, data2 and data3 (magnified). Fig. 11 11 data1 data3 GPS 2 GPS measurement result of data1, data2 and data3 (magnified, no.2). 15 [m] 9 10 ICP DP 10 7.0 [m] 9 GPS 11 data1 data3 GPS GPS Table 2 2 Processing times for integration process. 1 [sec] 2 [sec] [sec] 1,159.640 1,154.875 2,314.515 (15 [m]) 0.506 0.494 1.000 (30 [m]) 0.515 0.497 1.012 (50 [m]) 0.518 0.494 1.012 (100 [m]) 0.512 0.490 1.002 2 ICP DP 1 2 2 2,314.515 [sec] 1.0 [sec]
266 18 [fps] 10 2 4.2 4.1 12 data4 data6 data4 data6 data1 data4 data6 data4 data5 data6 13 13 ( 50480, 145050) GPS 12 data4 data6 Fig. 12 Road shapes of data4, data5 and data6. 13 data4 data6 GPS Fig. 13 GPS measurement result of data4, data5 and data6.
267 12 data5 data6 13 1 GPS 2 5. GPS RANSAC DP 1) (2006). 2) Google http://www.google.co.jp/help/maps/streetview/behind-the-scenes.html ( 2010-03-31). 3) http://www.premium-club.jp/technology/tech1.html ( 2010-03-31). 4) G-BOOK mx G-BOOK http://g-book.com/pc/whats G-BOOK mx/technology/probe.asp ( 2010-03-31). 5) G-BOOK mx G-BOOK http://g-book.com/pc/whats G-BOOK mx/technology/map.asp ( 2010-03- 31). 6) MIRU2009 IS3-30, pp.1570 1577 (2009). 7) D Vol.J92-D, No.2, pp.215 225 (2009). 8) ITS2004-68, pp.19 23 (2005). 9) GPS 8 IEEE 8th HISS pp.247 248 (2006). 10) ITS2006-62, pp.109 114 (2007). 11) ITS ITS http://internetits.org/ja/projects/pdf/report.pdf ( 2010-03-31). 12) Besl, P.J. and McKay, N.D.: A Method for Registration of 3-D Shapes, IEEE Trans. PAMI, Vol.14, No.2, pp.239 256 (1992). 13) Fischler, M. and Bolles, R.: Random sample consensus: A paradigm for model fitting with applications to image analysis and automated cartography, Graphics and Image Processing, Vol.24, No.6, pp.381 395 (1981). 14) DP
268 PRMU2006-166, pp.31 36 (2006). ( 22 3 31 ) ( 22 10 4 ) 1993 1995 2006 2007 ITS IEEE 2008 2010 ITS 1983 1988 1993 2007 OR