Bioimpedance technology (BMI) is a non-invasive detection technology that widely applied in volume measurement, human tissue structure analysis and human body composition analysis. It uses the electrical characteristics, such as impedance, admittance and dielectric constant, and changes of biological tissues and organs to extract biomedical information about human physiological and pathological conditions. This paper analyzes the influence of the distribution of frequency points on the characteristic parameters in the measurement of bio-impedance spectrum. A new method of sampling frequency points, namely the real axis equidistant method, is proposed, and this algorithm is used in the actual test. Without changing the total number of measurement frequency points, this method can effectively improve the accuracy of feature parameter α and fc calculation by increasing the number of intermediate sampling frequency points, and improve the accuracy of human body composition estimation. At last, the impedance of human body section is measured, and the characteristic parameters of human body section bio-impedance spectrum are analyzed. The continuous bioelectrical impedance changes caused by water and sugar were observed. It was initially found that drinking water led to an increase in impedance and drinking sugar led to a decrease in impedance. It lays a theoretical foundation for further research on human body data.