Towards automatic internal quality grading of mud crabs, a spectrometric analysis has been conducted in the visible and near-infrared range for essential components (shell, ovary, and meat) of a mud crab (Scylla paramamosain), which is an important product in the Mekong Delta region. Since the internal qualities of mud crabs, such as meat yield and ovarian fullness, are manually evaluated and graded at present, it is strongly desired to automate them by a nondestructive and quantitative technique. Spectrometry is one of the promising techniques, but it is difficult to develop a practical system because of the complexity of in-vitro study and consequent difficulties in dataset preparation for adopting machine-learning techniques. In this study, we proposed a research strategy of cumulative spectrometric analysis by developing various spectrometric systems optimized for three stages of in-vitro, semi-in-vivo and in-vitro conditions to enable an effective dataset preparation for developing internal quality models with machine-learning techniques. A preliminary experiment under in-vitro condition showed that the transmittance of the crab essential components revealed different spectrometric characteristics in the near-infrared wavelength region (at 940, 760, and 680 nm for shell, ovary, and meat, respectively). This result suggests that spectrometric analysis has a great potential for internal quality grading of mud crabs.