# US 11,026,571 B2 · Method for processing pupil tracking image > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,026,571 B2 - Document ID: US11026571B2 - Status: Granted - Kind: B2 - Application number: 16/469,587 - Filed: 2018-12-13 - Published: 2020-10-01 - Granted: 2021-06-08 - Earliest claimed: 2018-11-12 - Assignees: Wenzhou Medical University - Applicants: Wenzhou Medical University - Inventors: Jinhai Huang; Hang Yu; Meixiao Shen; Hao Chen - Web page: https://patentgrep.com/patent/US11026571B2 ## CPC - A61B 3/0025: {characterised by electronic signal processing, e.g. eye models} - G06T 7/246: using feature-based methods, e.g. the tracking of corners or segments - A61B 3/0008: {provided with illuminating means (A61B3/117, A61B3/14 take precedence)} - A61B 3/112: {for measuring diameter of pupils} - G06T 7/0012: {Biomedical image inspection} - G06T 7/60: Analysis of geometric attributes - G06T 2207/10016: Video; Image sequence - G06T 2207/30041: Eye; Retina; Ophthalmic - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A method for processing pupil tracking images, comprising the steps of: 1) acquiring eye images multiple times; 2) processing each acquired eye image, and determining whether a pupil can be positioned in the acquired image; and 3) determining that pupil tracking is completed when the pupil can be positioned in consecutive n pieces of eye images. Image processing efficiency is improved by searching a potential bright spot area and carrying out Blob analysis smearing images caused by pupil movement can be eliminated, the pupil edge is accurately positioned by adopting a line drawing method in collaboration with a support vector machine when determining the pupil edge, and finally a weighted least square method is adopted to fit a circle/ellipse (pupil). Therefore, real-time tracking and analysis of the pupil under a complex background are realized, and reliable data are provided for subsequent steps to obtain the refractive parameters of the eye. ## Claims ### Claim 1 A method for processing pupil tracking images, comprising the steps of: 1) acquiring multiple eye images; 2) processing each acquired eye image, and determining whether a pupil can be positioned in the acquired image; and 3) determining that the pupil tracking is completed when the pupil can be positioned in multiple consecutive eye images of the multiple eye images; wherein in step 2), the processing method for each acquired eye image includes the steps of: step 1, searching bright spots formed by projection of illumination light on a pupil; step 2, performing Blob analysis on the bright spots to determine a potential pupil area; step 3, carrying out gray normalization on the potential pupil area and carrying out multi-scale binarization; step 4, finding a potential circular area through Blob analysis, and obtaining a gravity center and a rough radius of the circular area; step 5, taking the gravity center obtained in the step 4 as the center of a circle, adopting a 360-degree outward line drawing method to obtain gray level jump, and obtaining the maximum gray level jump of the left, right and lower parts of the pupil, namely the maximum jump being the boundaries of the left, right and lower parts of the pupil; step 6, obtaining a boundary and determining the reliability of the boundary by adopting a support vector machine (SVM) according to a gray value obtained by a line drawing method, the boundary being an upper edge area of the pupil; and step 7, according to the boundaries of the left, right, lower and upper parts of the pupil obtained in step 5 and step 6, the pupil is obtained by using a least square method to fit a circle. ### Claim 2 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 5, the relationship between the gray level on each drawn line and the distance is obtained. ### Claim 3 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 2, a pupil image obtained by eccentric photography optometry is examined, and whether the pupil moves or not is determined by judging a shape of the light spot. ### Claim 4 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 3, a multi-scale binarization threshold value is determined according to the gray mean value, the gray standard deviation, the gray minimum value and the gray maximum value of the pupil image, wherein the binarization threshold value=gray mean value-α*gray standard deviation, and the coefficient α is an empirical value. ### Claim 5 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 6, a boundary is obtained by using a support vector machine (SVM) through the gray value sequence obtained by the 360-degree outward line drawing method in step 5. ### Claim 6 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 5, the maximum jump of the gray level is obtained through a difference method. ### Claim 7 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 7, a parameter equation of the circle is established as follows: x=z1+r cosφ y=z2+r sinφ wherein (z1, z2) is the coordinate of the circle center, r is a radius of the circle, φ is a parameter, and each point on the circle is T1=(x(φi), y(φi)), for an image pixel coordinate Pi=(xi1, xi2) required to be fitted with the circle, a least square method is used to fit distances form all the points Pi to points Ti, and minimization is carried out, d i 2 = min φ i ⁢ [ ( x i ⁢ ⁢ 1 - x ⁡ ( φ i ) ) 2 + ( x i ⁢ ⁢ 2 - y ⁡ ( φ i ) ) 2 ] , ∑ i = 1 m ⁢ d i 2 = min , further, in order to obtain the contribution of different pixel coordinates to the circle fitting, the minimization may be made as ∑ i = 1 m ⁢ w i * d i 2 = min , Wi is a weight coefficient. ### Claim 8 (depends on 1) The method for processing pupil tracking images according to claim 1, wherein in step 3), a number of multiple consecutive eye images of the multiple eye images is six consecutive eye images.