Due to the traditional contact measurement method which needs to contact with components and the limitations such as measurement time and human factor, a large-scale component measurement system based on machine vision has been developed to meet the non-contact, fast measurement speed, high measurement precision, real-time display and other measurement requirements. Firstly, in the large-scale component measurement system based on machine vision, the collected images are pretreated. Wavelet denoising method is used to finish the smooth denoising of the images in view of the existence of noise and other factors. Secondly, the scale invariant feature transformation (SIFT) algorithm is selected for image registration, the weighted average fusion algorithm based on the weighted average of the hat function is used to complete the image fusion without gaps, thus image mosaic is completed. And then, an improved single pixel edge detection method is proposed to extract the edge contour of the component image based on the Canny detection operator. Finally, geometrical experiment measurement is performed to the edge contour of the extracted images and the experimental data are analyzed. The large-scale component measurement system based on machine vision has simple operation, inexpensive price and the size of the component can be measured quickly within the allowable range of error.