The invention relates to the technical field of image processing, in particular to an image fusion algorithm based on textural features, which comprises the following steps: S1, judging whether an image acquired by a visible light sensor needs enhancement processing or not, performing enhancement processing if the light is dark, and otherwise, not performing enhancement processing; s2, acquiring abase layer of the image obtained in the step S1; s3, subtracting the base layer image from the original image to obtain a detail layer of a bright pixel, similarly, subtracting the original image from the base layer image to obtain a detail layer of a dark pixel, and if the image is an invisible light image, removing halo in the detail image by using halo edges to obtain a final detail image after halo removal; s4, fusing the base layers of the images acquired by the sensors in a weighted manner; and S5, performing weighted addition fusion on the base layer fusion image obtained in the step S4 and a bright pixel detail layer, and performing weighted subtraction fusion on the fused image and a dark detail layer to obtain a final fusion image. The algorithm provided by the invention can clearly reflect the position and scene details of the object.