| Summary: | 碩士 === 國立臺灣海洋大學 === 水產養殖學系 === 101 === In this experiment, blood parrot fish (Cichlasoma citrinellum × C. synspilum) as a model for conducting color grade in color machine vision system test (CMVS). There were four experiment trials in the study.
Experiment one : To identify the male and female in blood parrot. The tested fish were fed three different diets (commercial feed A, astaxanthin diet, astaxanthin and 4ppm methyl- testosterone diet) for 50 days and then fed basal diet (no pigment and methyltestosterone) for 5 months. It was observed that the genital pore shape (flake) could distinguish male or female at the accuracy was 97%. Based the above method to distinguish male or female, it was measured respective L*a*b* values, and in relationship between total weight and standard length. The result indicated that the fish fed with methyltestosterone did not show significant difference between the male and female in length and a* value.
Experiment two: To find a suitable environmental conditions for shooting with color machine vision system (CMVS). In this study, five different lighting combinations (including all shooting plate light, side light shooting plate with both sides, no shooting board with full light, no shooting board containing sidelight on both sides, no shooting board with xiece light) and 3 different colors plate (blue, black, white) were tested by using Video TesT – Master Morphology V4.0 to analyze the photographic value. The result show that white plate in shooting didn’t affect off-color and shooting board with full light appear without reflection.
Experiment three: Compare three different video test software (Video TesT, Photoshop, and Image Pro-plus) to find better software which result is close to human eyes in grading color grade of ornamental fish. The three software showed a similar result for grading color as compare to human eyes. However, only Video TesT show the most convince merit in setting parameters for repeat measurement.
Experiment four: To build a system for grading color of blood parrot fish. In this study, the tested fish were graded to 46 body color level and selected 100%, 410 single color level (from body color level) and selected 100%, 410 single color level and selected 70%, total 3 treatments. The above design was conducted by referencing previous three experiment results. The best result obtained at treatment of 410 single color level and selected 70%. The treatment was used to compare grading fish either on shooting board or in aquarium with water and the result was similar. It indicates blood parrot fish can be graded in aquarium to decrease stress caused by moving fish out of water for color grading.
|
|---|
