Research on wall painting-assisted copying algorithm based on homography matrix

CHENG Xiaoqi; ZENG Zhihao; LIU Zhimin; WENG Xiangwen

Abstract：In response to the challenges of extensive wall painting operations,such as the lack of flexible observation and tracing methods in manual painting processes,the complexity of alignment in projection-based painting methods,and the high cost of purchasing wall painting equipment,this paper proposes a wall painting assistance tracing algorithm based on homography matrix distortion correction.The algorithm first uses a planar target to calibrate the monocular camera to achieve distortion-free image acquisition;then,the planar target is affixed to the wall painting area,ensuring that the wall painting area is completely within the camera's field of view,serving as a positional constraint for camera placement.Feature extraction algorithms are then used to accurately extract feature points.Finally,a linear transformation algorithm is employed to estimate the homography matrix between the wall painting plane and the frontal plane,deriving the frontal view of the wall painting area and performing weighted fusion with the image to be drawn,allowing real-time observation of the fusion effect to achieve the tracing painting function.The accuracy experiment compares the frontal views obtained from different shooting positions to verify the effectiveness and accuracy of the proposed wall painting assistance tracing algorithm.Practical experiments using low-resolution cameras for physical verification show high fidelity,demonstrating the flexibility of this method.