Automatic circle pattern extraction and camera calibration using fast adaptive binarization and plane homography

Abstract

A method to detect calibration patterns using fast adaptive binarization and plane homography for images having complicated backgrounds and taken under lighting condition of industrial fields is proposed. The preprocessing step of involving the calibration of a camera, as required to measure object dimensions, must be able to extract calibration points from a calibration pattern. However, proper lighting conditions for the camera calibration of a measurement system are rarely provided in industrial settings. In this paper, a plane homography of calibration points extracted by fast adaptive image binarization is used for the automatic detection of a local set of calibration points. The integral image method is used as a central part for the rapid processing of adaptive image binarization. The extracted point set provides a transformation matrix of the plane homography and guides the extraction of other calibration points for cluttered backgrounds with irregular lighting conditions. During the calibration of the camera, a calibration point filtering procedure by a data In/Out system iteratively selects more correct calibration points while rejecting outliers and noisy data. Experiments using actual images verify that the proposed method can be applied to camera calibration with poor quality images having cluttered backgrounds obtained under uneven illumination.

Fast and Easy Method of Automatic Camera Calibration

“…Camera calibration is to gain the camera intrinsic parameters and the external parameters of the conversions between the camera and target by the maths conversion and optimizing method on the image model [9]. And the imaging model stands the relation of the spatial point and the pixel coordinate, and involves four coordinate systems, such as World coordinate system   , , , …”

An Automatic Calibration Method for AVM Cameras

“…To suppress misrecognition, binarization is used to verify whether the pattern is circular . The proposed method uses binarization to find the outer part of the circle by separating the circle and background parts, and to verify the pattern circularity by detecting the direction of the edge in the outer part of the circle [19]. Figure 8 depicts the result of detecting a circular pattern in an input image.…”

Accurate invariant pattern recognition for perspective camera model

“…[3][4][5][6] Only accurate image processing techniques can provide robust results on estimation an object position. [7][8][9][10][11][12] But image perspective deformation ( fig. 1) can significantly reduce recognition performance of known techniques and can lead to mistakes in pattern position estimation.…”