The invention discloses a pupil detection method based on edge filtering, ellipse evaluation and pupil verification, and the method mainly comprises the steps: firstly, carrying out the normalizationof an input gray level image, carrying out the Canny edge filtering of an eye image, and removing the noise which may damage the peripheral edge of a pupil; based on the physiological characteristicsof the eye image, through the linearity, internal intensity values and elliptical features, carrying out ellipse fitting and selecting an optimal ellipse and a pupil verification check to collect andevaluate the edge of the connection; and if the effective ellipse describing the pupil is found, returning the result, otherwise, performing the next calculation, reducing the image, and after applying the surface difference filter and the mean value filter to the re-scaled image, selecting the optimal position through the product result of the two filters and selecting the maximum value; and scaling the pixel position in the image results in a distance error of the pupil center in the original image. Thus, it must be necessary to optimize the pupil position on the original image based on theanalysis of the surrounding pixels of the selected pupil position.