Edge Localization Methods in Iris Recognition Systems

In iris recognition systems, precise edge localization serves as a critical step for ensuring identification accuracy. The iris boundaries typically manifest as two approximately circular contours—the pupillary boundary and the outer iris boundary. Achieving high-precision edge localization requires integrating edge detection techniques with geometric fitting methods.

The edge tracking localization method functions as a preliminary processing stage by analyzing image grayscale variations to identify potential boundary points. This approach initiates from candidate points and traces continuous edge points along grayscale gradient directions, effectively mitigating noise interference. During tracking, adaptive thresholding strategies are employed to ensure stable detection under varying illumination conditions. Implementation typically involves gradient magnitude calculation using operators like Sobel or Canny, followed by directional tracking algorithms that maintain connectivity through pixel neighborhood analysis.

After obtaining edge point sets, the least-squares circle fitting method comes into play. This mathematical optimization technique calculates the optimal circular parameters (center coordinates and radius) that best match the discrete edge points. Its advantage lies in effectively eliminating the influence of outlier points while providing an optimal circle equation that fits the majority of edge points. In practical applications, it's often combined with RANSAC (Random Sample Consensus) algorithm to enhance noise resistance. Code implementation would involve matrix operations for solving the circle equation: [x² + y² + Dx + Ey + F = 0] using singular value decomposition or linear algebra solvers.

Visual diagrams typically demonstrate three key phases: original iris images showing blurred annular structures; edge tracking results highlighting boundary contours as scattered points; and finally, two fitted concentric circles precisely superimposed on the inner and outer iris edges. This processing pipeline—from coarse positioning to refined correction—establishes the geometric foundation for subsequent feature extraction. Programmatically, this would involve OpenCV functions like HoughCircles or custom implementations combining contour detection with iterative fitting routines.