A method and device for coding an image of a video sequence by image block using the time correlation between images based on movement field calculation. The method includes subdividing the images into blocks of images and a step for configuring the blocks into regions. A region corresponds to a succession of consecutive or non-consecutive blocks, according to the television scanning order of the image. In the inter- or intra-block coding step, the blocks are coded by region according to an ordering of the regions and a region according to the order of succession of the blocks that make up the region. The coded block error correction step includes analyzing the field of movement, a configuring step, and an error correction step. The step for analyzing the field of movement is carried out to calculate a “lifespan” parameter assigned to a block and corresponding to its probability of existence in subsequent images, according to the movement vector associated with this block. The configuring step is performed taking into account the value of this parameter to define and order the regions of the image from most important to least important. The error correction step uses a more effective error correction algorithm for the coded data corresponding to the most important regions than for the data corresponding to the least important regions.