Abstract: A multi-sensor fusion detection technology for elliptical glass curtain wall defects based on improved least squares method was proposed to address the issue of significant errors in the detection results of conventional multi-sensor fusion detection for elliptical glass curtain wall defects. After analyzing the stress value and contrast enhancement of the grouped sensor detection data signal at the force point of the elliptical glass curtain wall to obtain the initial fitting curve of the defect location, the improved least squares method was used to fit the three-dimensional position curve of the defect location, and the defect detection images were fused to obtain the detection results. The experimental results demonstrated that the defect detection results of elliptical glass curtain walls obtained by the proposed method showed an average detection accumulative error of 0.07, which can meet the defect detection requirements of elliptical glass curtain walls.