한수지 48(2), 221-232, 2015 Original Article Korean J Fish Aquat Sci 48(2),221-232,2015 고등어 (Scomber japonicus), 불볼락 (Sebastes thompsoni) 및쥐노래미 (Hexagrammos otakii) 에의한광대역음향산란신호의시간 - 주파수분석 이대재 * 부경대학교해양생산시스템관리학부 Time-Frequency Analysis of Broadband Acoustic Scattering from Chub Mackerel Scomber japonicus, Goldeye Rockfish Sebastes thompsoni, and Fat Greenling Hexagrammos otakii Dae-Jae Lee* Division of Marine Production System Management, Pukyong National University, Busan 608-737, Korea Broadband echoes measured in live chub mackerel Scomber japonicus, goldeye rockfish Sebastes thompsoni, and fat greenling Hexagrammos otakii with different morphologies and internal characteristics were analyzed in time and frequency domains to understand the species-specific echo feature characteristics for classifying fish species. The mean echo image for each time-frequency representation dataset obtained as a function of orientation angle was extracted to mitigate the effect of fish orientation on acoustic scattering. The joint time-frequency content of the broadband echo signals was obtained using the smoothed pseudo-wigner-ville distribution (SPWVD). The SPWVDs were analyzed for each echo signature of the three fish species. The results show that the time-frequency analysis provided species-specific echo structure patterns and metrics of the broadband acoustic signals to facilitate fish species classification. Key words: Time-frequency analysis, Spectrogram, SPWVD, Broadband acoustic scattering, Echo signature 서론,,, (Foote, 1980; Clay and Horne, 1994; Traykoyski et al., 1998; Jaffe, 2006; Nesse et at., 2009; Stanton et al., 2010).,,,, (non-stationary). chirp echo, (Fourier transform)., -,., - STFT (short time frequency transform), WVD (Wigner-Ville distribution), SPWVD (smoothed pseudo Wigner-Ville distribution) (Rihaczek, 1968; Imberger and Boashash, 1986; Safizadeh et al., 2001; Shui et al., 2007; Dong and Cui, 2012). SPWVD chirp 2 (power spectral density), http://dx.doi.org/10.5657/kfas.2015.0221 Korean J Fish Aquat Sci 48(2) 221-232, April 2015 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial Licens (http://creativecommons.org/licenses/by-nc/3.0/)which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Received 19 February 2015; Accepted 10 April 2015 *Corresponding author: Tel: +82. 51. 629. 5889 Fax: +82. 51. 629. 5885 E-mail address: daejael@pknu.ac.kr Copyright 2014 The Korean Society of Fisheries and Aquatic Science 221 pissn:0374-8111, eissn:2287-8815
222 이대재 spectrogram, 2., Hanning chirp echo (segment)., chirp echo SPWVD - pseudo-colored image (pcolor),. STFT spectrogram ( ) (trade-off). SPWVD, WVD (crossterm interference).,, chirp, chirp echo, SPWVD - echo.,,. 재료및방법 어류의자세변화에따른광대역음향산란신호의변동특성측정 Lee et al. (2015) chirp, 3 chirp echo, -,, Fig. 1.,., (single chamber),, (double chamber) echo, echo,., echo,,. Fig. 1 - echo 25 2.5,., 25 21 - echo. Fig. 1. (a) Ray diagram for eleven angles of orientation for measuring the broadband echoes from live fish individuals. (b) The lateral and ventral radiographs for fish (goldeye rockfish Sebastes thompsoni) to be measured. The air-filled swimbladder, with double chambers, and the backbone, including the spines and ribs, can be seen as the dark and white shapes, respectively. 어류에의한광대역음향산란신호의시간 - 주파수분석,,., (negative),., s(t) Hilbert, x(t) (analytic signal) -., s(t) Hilbert H[s(t)], x(t) x(t) = s(t)+jh[s(t)] (Safizadeh et al., 2001). (positive). SPWVD, 1 MHz, bin 128, (smoothing window) Hanning
광대역음향산란신호의시간 - 주파수분석 223, (N ) N=64. Hanning Hamming (edges),, N (Schilling and Harris, 2000). STFT (1) (McClellan et al., 1998; Costa et al., 2012; Han and Kim, 2010; Gavrovska et al., 2010). STFT x (f, ) =X(f, ) = + - x(t) w*(t- ) e-j2 fn dt (1) x(t), t (angular frequency, =2 f )., w(t- ) t =., (1) STFT (discrete), STFT x (f, ) = X(f,m) = x(n) w(n-m) e-j2 fn (2) n=-, m n t index, x(n) echo, w(n)., x(t), spectrogram (3). S x (f,m) = X(f,m) 2 (3), (2) w(n) w(n) = 0.5 (1-cos( 2πn )) (4) N-1 Hanning, N. STFT -, ( ),,.., STFT SPWVD -. chirp echo,,. WVD STFT, chirp echo. chirp echo SPWVD, SPWVD STFT - echo,. WVD SPWVD (Blaska and Sedlacek, 2001; Dong and Cui, 2012) SPWVD x (f,t) = g(t)*( + - h( )[x(t + )x * (t - )]e -j d ) (5) 2 2., * t (convolution), g(t) (time smoothing window function), h( ) (frequency smoothing window function). (5), SPWVD x (f,t)= + - h( ) + - g(s-t)[x(s+ )x * (s- )]ds e -j d 2 2 (6), (Shui et al., 2007), SPWVD x (f,n)= h(m) g(k) x(n+m+k) x* (n+m-k) e -j4 fk n=- n=-. m k t index, x(n). (7) -, SPWVD g(k) h(m),. chirp chirp sweep (T ) (B) - ( ) (8) (Rihaczek, 1968; Imberger and Boashash, 1986.). = 2 df (t) i -1/2 = dt 2T B (8) ( bin ), f i (t) (instantaneous frequency), f i (t)/dt. 시간 - 주파수 echo 응답패턴으로부터어종식별정보 (7) (8)
224 이대재 의추출 chirp echo - echo Fig. 2. Fig. 2 echo 25 2.5 21 -., 2,, image.,,, echo signature. Fig. 2 -, sweep, (maximum instantaneous bandwidth),,, echo (duration),,. 결과및고찰 STFT 와 SPWVD 기법에의한 chirp echo 신호의시간 - 주파수응답특성 chirp echo STFT SPWVD - echo Fig. 3. Fig. 3 (a) -,, spectrogram., echo smearing.,, echo smearing Fig. 2. Parameters of the echo feature extracted from the timefrequency representation showing the species-dependent characteristics of fish.., spectrogram, (trade-off). 0.3 ms, 220 khz (75-295 khz) chirp, FOM (figure of merit) 90 khz 230 khz echo. chirp., FOM sweep 0.23 ms 140 khz (90-230 khz), Hanning ( ) (8), =57. -, chirp = 64. Fig. 3 (a) (b) ( = 64) STFT SPWVD - echo. Fig. 3 STFT SPWVD echo,, SPWVD STFT, SPWVD chirp echo -. 고등어, 불볼락, 쥐노래미에대한어종식별정보의추출 chirp Fig. 1-25 +25 chirp echo 2.5 Fig. 4. Fig. 4 (V), (ms), 1~21., 1-25 (head-down), 11 0 (normal aspect), 21 +25 (head-up). X-ray 11.9, Fig. 4, (normal aspect) 16. Fig. 4 16,. Fig. 4 chirp echo,,, echo (Foote, 1980)., contrast echo
광대역음향산란신호의시간 - 주파수분석 225 Fig. 3. Time-frequency representations (TFRs) of the broadband echoes from goldeye rockfish Sebastes thompsoni. (a) Short time Fourier transform (Hanning window, h=62 points), (b) Smoothed pseudo Wigner-Ville distribution (Hanning window, g=h=62 points).. Fig. 4., Fig. 4 chirp echo ( ) peak null, 16 peak null. echo Fig. 4 chirp echo SPWVD - Fig. 5. Fig. 5 (khz), (ms), 1~21 Fig. 4. Fig. 5 (in phase)
226 이대재 Fig. 4. Chirp echo waveforms measured as a function of angle of orientation in the dorsal plane of a chub mackerel Scomber japonicus. The strongest echoes occurred when the incident signal was perpendicular to the long axis of the swimbladder (No. 16). The echoes were measured at increments of 2.5 over an angle range of -25 (head-down, No. 21) to +25 (head-up, No. 1). echo. echo, (out of phase) echo. echo peak null - echo. Fig. 5 16, 13 ~17 - echo, echo peak null. chirp chirp,, FOM Fig. 6. Fig. 6 (khz), (ms), 0~10 V. Fig. 6 FOM - echo chirp - echo. Fig. 6 bin (Nx) bin (N) Nx=512 N=128,, (time smoothing window, g) (frequency
광대역음향산란신호의시간 - 주파수분석 227 Fig. 5. Time-frequency representations for the broadband echoes measured over a range of -25 (No. 21) to +25 (No. 1) with 2.5 steps in the dorsal plane of a chub mackerel Scomber japonicus. The echoes showed the highest amplitude and simplest signature when the incident angle was perpendicular to the long axis of the swimbladder (No. 16). The complexity of the echoes in both the time and frequency domains increased as the aspect angle of the swimbladder moved away from the normal aspect (No. 16). smoothing window, h) bin g=h=64. Fig. 6 FOM 2 V, Fig. 2 sweep 117.2 khz, 0.183 ms, 3.5 V sweep 82.03 khz 0.121 ms, 35.16 khz,, echo [0.277 ms, 148.44 khz]., Fig. 5-25 +25 21 - - echo, Fig. 7. Fig. 7 (a1), (a2), (a3),, chirp echo -., (b1), (b2), (b3) (a1), (a2), (a3) Fig. 6 FOM., chirp - echo ( -.). Fig. 7 - Fig. 2
228 이대재 Fig. 6. Time-frequency image for the calibrated broadband signal of a chirp echo sounder. The spectrogram was calculated using the smoothed pseudo Wigner-Ville Distribution with a Hanning window (time and frequency smoothing, g=h=64 points and frequency bins N=128 points).,,, sweep,,,, echo 3.5 V,., Fig. 7 (a1), echo 0.198 ms 0.309 ms ( 0.111 ms), 97.66 khz 167.97 khz (sweep 70.31 khz), 42.97 khz, echo [0.253 ms, 128.91 khz], echo [0.270 ms, 152.34 khz]. Fig. 7(a1) echo,., echo null echo,. Fig. 7(a2) echo 0.197 ms 0.291 ms ( 0.094 ms), 101.56 khz 164.06 khz (sweep 62.50 khz), 35.16 khz, echo 1 2 [0.225 ms, 117.19 khz] [0.253 ms, 140.63 khz], echo [0.236 ms, 128.91 khz]., Fig. 7(a3), echo 0.209 ms 0.273 ms ( Fig. 7. Time-frequency response patterns of broadband echoes from chub mackerel Scomber japonicus (a1), goldeye rockfish Sebastes thompsoni (a2) and fat greenling Hexagrammos otakii (a3) before (a1, a2, a3) and after removing (b1, b2, b3) the effect of transmitting and receiving characteristics of chirp echo sounder. The regions of x and z indicate the destructive and constructive effects.
광대역음향산란신호의시간 - 주파수분석 229 0.064 ms), 125.0 khz 167.97 khz (sweep 42.97 khz), 31.25 khz, echo [0.235 ms, 144.53 khz], echo [0.197 ms, 125.0 khz]. echo sweep,,,., Fig. 6 FOM Fig. 7 (a1), (a2), (a3) -,, Fig. 7 (b1), (b2), (b3) (masking) 1.5 V., Fig. 6 FOM Fig. 7 (a1), (a2), (a3) echo 1.5 V., echo. Fig. 7 (b1), (b2), (b3), echo FOM. chirp,,.,., (b1), FOM [0.183 ms, 109.38 khz] [0.313 ms, 195.31 khz],, [0.247 ms, 105.47 khz] [0.345 ms, 179.69 khz] echo., echo., FOM [0.197 ms, 109.38 khz] [0.249 ms, 160.16 khz],, [0.245 ms, 109.38 khz] [0.339 ms, 179.69 khz]., echo., FOM [0.156 ms, 109.38 khz] echo, [0.186 ms, 132.81 khz] [0.284 ms, 140.63 khz],, [0.222 ms, 109.38 khz] [0.295 ms, 152.34 khz].,,, echo echo,. Fig. 7 (a1), (a2), (a3) echo Fig. 8. Fig. 8 (a) echo, (b) echo,, (ms) (khz)., Fig. 8 (a),, peak 0.245 ms, 0.236 ms, 0.236 ms, 67.4, 66.0, 59.8., echo 0.125~0.19 ms, peak,,., 0.27 ms, 0.30 ms, peak 0.111 ms, 0.102 ms, 0.089 ms,,,. Fig. 8. Comparison of the echo response patterns for chub mackerel Scomber japonicus, goldeye rockfish Sebastes thompsoni and fat greenling Hexagrammos otakii in the time (a) and frequency domains (b).
230 이대재, Fig. 8 (b),, 2 peak, 1 peak 132.81 khz, 140.63 khz, 144.53 khz,, 2 peak 117.19 khz, 113.28 khz, 113.28 khz. 1 2 433.24, 428.61, 369.16 419.56, 354.55, 214.82, null, 121.09 khz., peak sweep 67.39 khz, 61.14 khz, 56.33 khz,,,. sweep,,. echo 0.24 ms, echo,,,. 140 khz 117 khz, 121.09 khz,.,, -25 +25 21 chirp echo - image Fig. 9. Fig. 9(a) 0~20 db. Fig. 9(b) Fig. 9(a) - 3, (as), (bs), (cs) echo,, echo. Fig. 9(c) Fig. 9(b) 3 dataset 2 dataset, PC1 1, PC2 2 Fig. 9. (a) Spectrograms for broadband echoes measured from live individuals of chub mackerel Scomber japonicus (a1), goldeye rockfish Sebastes thompsoni (a2) and fat greenling Hexagrammos otakii (a3). (b) Scatter plot of the time-frequency dataset (as: chub mackerel, bs: goldeye rockfish, cs: fat greenling) in 3D space, (c) Projections of the points of the dataset on the PC1, PC2 (first and second principal components) plane of 2D space. The spectrograms are indicated as a relative level. The F1, F2 and F3 indicate the locations of center points.
광대역음향산란신호의시간 - 주파수분석 231. Fig. 9 (a) (b),, 2 - color contour map 3 -, 12~20 db color contour,,, -. -. 1,,, 2 1. 3 3 echo 1 1 2 2 scatter map., 1 2 Fig. 9(b) (,, ) 99.87%. Fig. 9(c) 2 scatter map - 1 (PC1) 2 (PC2),, [267.46, 5.30], [262.25, 5.96] [231.27, 8.25], (Euclidean distance) 5.26, 31.07, 31.07., 5, 3 -.,., Fig. 9 (a) (b) - (a) contour map, (b), echo contour,.,,, 1:0.87:0.60., (Lee et al., 2015) 13.7 cm 3 11.7 cm 3, 1: 0.854, -. - echo contour, echo., chirp - echo,. 사사 2013 ( ) ( NRF-2013R1A1A2004378). References Blaska J and Sedlacek M. 2001. Use of the intergral transforms for estimation of instantaneous frequency. Measur Sci Rev 1, 169-172. Clay CS and Horne JK. 1994. Acoustic models of fish: The Atlantic cod (Gadus morhua). J Acoust Soc Am 96, 1161-1668. Costa J, Ortigueira M, Batista A and Paiva T. 2012. Slip spindles detection: a mixed method using STFT and WMSD. Intern J Bioelectrom 14, 229-233. Dong Y and Cui Y. 2012. Analysis of a new joint time-frequency distribution of suppressing cross-term. Res J Appl Sci Eng Technol 4, 1580-1584. Fassler SMM, Fernandes PG, Semple SIK and Brierley AS. 2009. Depthe-dependent swimbladder compression in herring Clupea haengus obserbed using magnetic resonance imaging. J Fish Bio 74, 296-303. http://dx.doi.org/ 10.1111/j.1095-8649.2008.02130.x. Foote KG. 1980. Importance of the swimbladder in acoustic scattering by fish: A comparison of gadoid and mackerel target strengths. J Acoust Soc Am 67, 2084-2089. Gavrovska AM, Paskas MP and Reljin IS. 2010. Determination of morphologically characteristics PCG segments from spectrogram image. Teflor J 2, 74-77. Han SK and Kim HT. 2010. Efficient radar target recognition using a combination of range profile and time-frequency analysis. Progress Electrom Res 108, 131-141. Imberger J and Boashash B. 1986. Application of the Wigner- Ville distribution to temperature gradient microstructure: A new technique to study small-scale variations. J Physic Oceanography 16, 1997-2012. Jaffe JS. 2006. Using multi-angle scattered sound to size fish swimbladders. ICES J Mar Sci 63, 1397-1404. http://dx.doi. org/10.1016/j.icesjms.2006.04.024. Lee DJ, Kang HY and Kwak MS. 2015. Analysis and classification of broadband acoustic echoes from individual live fish using the pulse compression technique. Kor J Fish Aquat Sci 48, 207-220. http://dx.doi.org/10.5657/kfas.2015.0207. McClellan JH, Schafer RW and Yoder MA. 1998. DSP FIRST: A Multimedia Approach. Prentice Hall, NJ, USA, 349-368. Nesse TL, Hobek H and Korneliussen RJ. 2009. Measurement of acoustic-scattering spectra from the whole and pars of Atlantic mackerel. ICES J Mar Sci 66, 1169-1175. http:// dx.doi.org/ 10.1093/icesjms/fsp087. Rihaczek AW. 1968. Signal energy distribution in time and fre-
232 이대재 quency. IEEE Trans Inform Theory IT-14, 369-374. Safizadeh MS, Lepine A, Forsyth DS and Fahr A. 2001. Time-frequency analysis of pulsed eddy current signals. J Nondestruct Evaluat 20, 73-86. http://dx.doi. org/10.1023/a:1012244208475. Schilling RJ and Harris SL. 2000. Applied Numerical Methods for Engineering. Brooks/Cole, CA, USA, 497-500. Shui PL, Shang HY and Zhao YB. 2007. Instantaneous frequency estimation based on directionally smoothed pseudo- Wegner-Ville distribution bank. IET Radar Sonar Navig 1, 317-325. http://dx.doi.org/10.1049/rsn:20060123. Stanton TK, Chu D, Jech JM and Irish JD. 2010. New broadband methods for resonance classification and high-resolution imagery of fish with swimbladders using a modified commercial broadband echosounder. ICES J Mar Sci 67, 365-378. http://dx.doi.org/10.1093/icesjms/fsp262. Traykovski LVM, O Driscoll RL and McGehee DE. 1998. Effect of orientation on broadband acoustic scattering of Atlantic krill Euphausia superb: Implications for inverting zooplankton spectral acoustic signatures for angle of orientation. J Acoust Soc Am 104, 2121-2135.