| Two themes are discussed in this thesis: the age and growth of rockfish (sebastes schlegeli) and the design of the computer-assisted detection system of otolith rings.In the first part, collected rockfish samples monthly by the inshore of Qingdao were measured biologically, then the otolithes were taken and dried. Taking the otolithes as the stuff of age identify, through margin increment analyses, we got the cycle of the otolith growth ring, and know that it is annually. Base on analyses new ring of the edge of the otolithes, we knew that translucent band is formed from June to Nov, and opaque band is from Dec. to the next May. A morphological analysis of the otolithes was made. According to the datum of body length and weight, we got the length-weight regression for sexes combined: BW=10-5BL3.1267. The analyses of otolithes' weight and age indicate that there is linear relationship between the two factors, the linear equation is: OMt=0. 0323t-0. 0163, it proves that we can research the rockfish's age and growth by the weight of otolith. There is significant linear relationship between otolith's radius, length, width and the body length. No significant difference between male and female rockfish's growth, we can describe whole growth with one equation, equation of body length-at-age is BLt=483. 7922 (1-e-0.2985t). We can see that K , compare to former research, is a little bigger.In the second part, expatiate several models pattern setting methods of identifying rings image of otolith image, and compared them. All models is base on the alternant changing of brightness, which is produced by setting transmission light on otolith. One-dimensional simple detect system is base on the change of gray level of otolith image along single line, draw an undee profile. Take the maximum and minimum projective point of peak and valley, which is formed along the sample line, asdividing point, take the distance between two projective point as the width between the growth rings, the system providing more or less reliable ring count estimation, it's strongpoint is fast and easy to be used, it's defect is easy to get stochastic effect, and possibly lead wrong growth ring mark. The thesis improves this method, increase the sample area gradually, which reduce the error aroused by stochastic effect. The main theory of deformable template is that new ring is formed on the base of former ring, and they have similar contours, take the edge of otolith image as template, we can control well the direction of ring detection, however, it is not applicable when there is significant difference between the ring and edge contours. Currently, two-dimensional growth ring detection is a good identify method, it mainly induct the direction by the partial gray level information, template step sides were modulated by the otolith edge contour in order to get better identify result. The thesis develop the otolith detection software, which fulfill One-dimensional method, and compare the detect effect before and after optimized. |
