1, 1,2, 1 Hammering Test with Image and Sound Signal Processing Atsushi YAMASHITA 1, Takahiro HARA 1,2 and Toru KANEKO 1 1 Department of Mechanical Engineering, Shizuoka University. 2 Mitsubishi Electric. In this paper, we propose a new method for a hammering test by using image and sound signal processing techniques. A method for discriminating a property of an object with the use of generated sound when striking it with a hammer is called a hammering test. However, this method depends on human experience and skills. In addition, if we perform this test over a wide area of objects, it is required to manually record hammering positions one by one. Therefore, this paper proposes a hammering test system consisting of two video cameras that can acquire image and sound signals of a hammering scene. The shape of the object is measured by the image signal processing from the result of 3-D measurement of each hammering position, and the thickness or material of the object is estimated by the sound signal processing in time and frequency domains. The validity of proposed method is shown through experiments. Key Words : Image processing, Sound signal processing, Hammering test, Stereo measurement, FFT 1. (1) (3) (4) (8) FEM (5) (7) 1 (2) 2004 6 20 04 (9) (13) 2005 6 23 1 ( 432-8561 3-5-1) 2 Email: yamashita@ieee.org 1
Fig. 1 Hammer Two video cameras Object Overview of hammering test. 2 2. 2 (14) 1. 2. 3. (8) 1. 2. 3. 1 3 2 1 3. 2 3 1 3 1 1 T s T s 2
Amplitude Amplitude y(t) C e-b(t-ts) Time (s) f Frequency (Hz) Fig. 2 An example of raw sound wave and its approximation curve. Fig. 3 An example of frequency spectrum with FFT. t T s y(t + t) y(t) > L 1, (1) y(t + t) > L 2, (2) y(t) y(t) t t Fig. 4 L 1 L 2 60dB f i i A i i 1/1000 N 3 2 3 1 y(t) B V f y(t) B V f B-V f y(t) = C e B(t Ts), (3) t > T s 2 B B ISODATA (15) 2 3 1 2 ISODATA (16) FFT ISODATA f n V f 3 B-V f V f = 1 N A s A 2 i ( f n i f ) 2, (4) 4 f = A 2 s = i=1 N i=1 N i=1 f i A 2 i A 2 s, (5) A 2 i, (6) 3 Cluster 1 Cluster 2 Cluster 3 Center Example of clustering in B-V f diagram. 3 3 B
Cluster 1 Cluster 2 Yellow Magenta 20 40 20 1 Fig. 5 P1 B1 Cyan Cluster 3 Center B Example of discrimination map. 10 10 Mark (a) Schematic. (b) Photo. Fig. 6 Overview of the hammer. 3 2 B-V f B-V f Fig. 7 Extraction of the hammer from images. 5 4. (a) Acquired image. (b) Enlarged image of hammer. Fig. 8 Extraction of hammer s marks. 3 3 2 2 3 1 3 6 3 3 T s T s 4 1 2 T s 3 1 B V f 4 2 T s 7 3 2 2 3 8 2 3 4
Cluster 2 600 600 Cluster 3 47 Cluster 1 Left image Right image 12 450 Cluster 1 (a) Top side. (b) Back side. 312 Fig. 11 Object I. 5. Left camera Fig. 10 Fig. 9 Corresponding point Stereo measurement. Object with curved surfaces. Right camera 720 480pixel 30frame/s 44.1kHz 1 16 5 1 11 3 1 312mm 2 2 47mm 3 3 3 12mm 9 12 3 FFT 13 FFT V f 4 3 100Hz 1KHz 3 3 3 10 14 1 4 2 3 15 4 4 4 4 M M2 4 7 8 M2 4 3 OpenGL 3 16 16(a) 16(b) 5
(a) Cluster 1. (a) Cluster 1. (b) Cluster 2. (b) Cluster 2. (c) Cluster 3. (c) Cluster 3. Fig. 12 Examples of raw sound wave (vertical axis: amplitude, horizontal axis: time). 3 Cluster 2 5 2 17 4 4 Cluster 1 Cluster 3 B Fig. 14 Discrimination map I. 4 Fig. 13 4 Fig. 15 3-D positions and colors of each hammering 18 point I. 1 2 3 4 6 Examples of frequency spectrum with FFT (vertical axis: amplitude, horizontal axis: frequency).
(a) Experimental result. (b) Actual one. Fig. 16 Result of hammering test for object I. (a) Experimental result. (b) Actual one. Fig. 20 Result of hammering test for object II. Plastic Styrene foam 100 100 100 Cork 100 100 Wood Fig. 19 (a) Schematic. Cluster 2 Fig. 18 Fig. 17 Cluster 1 Object II. Cluster 3 Cluster 4 (b) Photo. B Discrimination map II. 3-D positions and colors of each hammering point II. 2 4 B V f 6 6 3 4 19 20 1 (3) C 2 3 5 3 Pentium IV 1.6GHz 1 3 2.3s 2.3s 2 3 1 1.06mm 2 4 3 4 96% 7
Table 1 Hammering power and parameter B, V f. Strong Medium Weak Faint B 42.8 39.3 42.3 34.5 V f 183 169 167 159 6. (2) M. J. Sansalon and W. B. Streett: Impact-Echo, Bullbrier Press, Ithaca, New York (1997) (3) :,, 57-10, pp.668 674 (2001) 3 (4) C. Cheng and M. Sansalone: The Impact-Echo Response of Concrete Plates Containg De-Laminations: Numerial, Experiential and Field Studies, Materials and Structures, 2 26, pp.274 285 (1993) (5), :,, 35-564, pp.169 176 (1997) (6),,, :,, 23-1, pp.589 594 (2001) 1. (7),,, :,, 55-704, pp.65 79 (2002) (8),,,, : 2. 2, 104, pp.65 69 (2001) (9) O. R. Gericke: Determination of the Geometry of Hidden 3. 2 Defects by Ultrasonic Pulse Analysis Testing, Journal of Acoustic Society of America, 35, pp.364 368 (1963) 4. (10) : 3,, 47-9, pp.636 641 (1998) 5. (11) T. Hirata and T. Uomoto: Detection of Ultrasonic Pulse Echo through Steel Bar in Concrete Crack 6. Depth Measurement, Non-Destructive Testing in Civil Engineering 2000, pp.383 390 (2000) (12) N. Kato, N. Saeki, Y. Tanigawa, K. Kato, T. Kawai and Y. Murata: Evaluation of Deterioration and Specification of Defect of RC Structure by Natural Potential and External Impact-Acoustic Methods, Materials Science Research International, 7-2, pp.138 146 (2001) (13),,, :,, 24-1, pp.1461 1466, (2002) (14),, :, C, 67-653, pp.17 22 (2001) FFT Wavelet (15) G. H. Ball and D. J. Hall: ISODATA - Novel Method of Data Analysis and Pattern Classification, Stanford Research Institute (1965) (16), : ISODATA, D-II, J82-D-II-4, pp.751 762 (1999) (17),, :, CVIM, 2004-40, pp.117 124 (2004) 8 (1),,, :, 04, 2A1-L2-11, pp.1 4, (2004)