A systemic calibration and a characterization approach were presented for acousto optic tunable filter based hyperspectral imaging system. A series of tests were conducted to validate the performance of a self-integrated hyperspectral imaging system. Results show that the presented approach effectively validate the systematic characteristics in both spectral and spatial domains. Through flat-fielding correction, the system responsivity non-uniformity, known as pattern noise, is apparently eliminated. The spectral sensitivity of the system is corrected by controlling both the camera exposure time and digital gains at each wavelength, which significantly improve the signal to noise ratio of the system. By studying the system outputs with different camera exposure time, instead of using an attenuator to adjust the input light intensity, a more efficient and cost effective approach to measure the spectral linearity of the system is presented. The proposed methods are also applicable to the liquid crystal tunable filter based hyperspectral imaging systems.