The invention discloses a video background extraction method based on a weighted sample. Step one, a video image is read in; step two, background modeling is performed on the read-in image, and the model is initialized if the read image is a first frame, or background detection is performed; step three, if the current pixel is a background pixel, background updating is randomly performed; and step four, a background image is acquired and the image is saved. According to the video background extraction method based on the weighted sample, sample updating is performed by using the weight of the sample and an "activity" sample so that error updating of effective samples can be reduced and the expression capability of the model can be enhanced. The use amount of the samples can be reduced and the calculation efficiency can be enhanced under the premise of guaranteeing the background detection accuracy so that the cost of video background extraction can be reduced, the real-time performance is better and the range of application is wide.