Machine vision fusion splicing method and system

The invention relates to the technical field of image processing, in particular to a machine vision fusion splicing method and system, and the method comprises the steps: inputting an input image into a Harris corner extraction model; the Harris corner extraction model calculates an autocorrelation matrix for each pixel point of the input image; performing Gaussian filtering on each pixel point of the input image to obtain a new autocorrelation matrix; calculating the angular point measurement of each pixel point based on the autocorrelation matrix and the new autocorrelation matrix; selecting a local maximum point from the angular point measurement; setting a threshold value, and if the local maximum value point is higher than the threshold value, determining that the local maximum value point is an angular point; and performing quantitative measurement by adopting the angular points to obtain feature points. And carrying out image registration based on the feature points, and establishing a geometric corres