Measurement of blood glucose concentration is important in the prevention and treatment of diabetes. To improve the measurement signal-to-noise ratio and accuracy, using a polarization camera as the sensor, this paper proposes a multi-wavelength dynamic optical rotation angle method to quantitatively analyze human blood glucose. First, this paper utilizes a split-time approach to drive nine different wavelengths of light sources to work separately to achieve the collection of multiple different wavelengths of data and increase the amount of information. Second, to capture images with an array sensor such as a camera, using space-for-accuracy, this paper uses the cumulative value of the gray values of all pixel points in the image at each angle as the light intensity value to improve the amplitude (light intensity) resolution. Third, the extraction method of the dynamic optical rotation angle in this paper is to extract the difference between the maximum and minimum values of the optical rotation angle from each cycle of the signal of the continuous cyclic change of the length of the optical path, and the accumulated value of the difference of multiple cycles is recorded as the dynamic optical rotation angle, which not only reduces the errors caused by the individual differences, the external environment and the measuring instrument, but also has the effect of averaging and improves the signal-to-noise ratio of the optical rotation angle. To prove the effectiveness of the proposed methods, human finger real and simulation experiments are carried out, respectively. Firstly, the polarization camera is used to collect the image data of elliptically polarized light of nine different wavelengths of linearly polarized light after passing through the human body’s finger, and obtains the optical rotation angle under each wavelength that changes with the photoplethysmography (PPG) through data processing, and extracts the dynamic optical rotation angles of the nine wavelengths by the extraction method, and the human body’s blood glucose can be quantitatively analyzed. Secondly, this paper builds a simulation model to simulate the blood volume conversion of the human finger, and uses the simulation experimental system to collect the dynamic optical rotation angles of nine wavelengths of a total of 137 human serum samples, and uses partial least squares regression to establish a model between the dynamic optical rotation angles and blood glucose concentration. In the results of predicting all samples using the model, the correlation coefficient and root mean square error were 0.7960 and 2.16 mmol/L, respectively, both of which improved by 137.26 % and decreased by 35.33 %, respectively, compared to the existing method. The experimental results show that the method in this paper has high accuracy in detecting the blood glucose concentration in human beings, which is of great significance in realizing the noninvasive quantitative analysis of human blood glucose components by the polarized light method.