The invention discloses a frame compression method for an image and a decompression method and device for the image. The frame compression method comprises the following steps of starting with the second row of pixel points of the image and carrying out the following operations line by line: determining a predicted pixel value of each pixel point of the current row according to the pixel values of the pixel points of the previous row; determining a difference between the predicted pixel value and the pixel value of each pixel point of the current row; determining a bit depth of each pixel point of the current row; grouping the pixel points of the current row and determining the bit lengths of each group of pixel points; and if the sum of the bit lengths of each group of pixel points is not greater than a bit length threshold, taking the bit depths and the differences of each group of pixel points as compressed data of the group of pixel points and packaging to a compressed data packet. According to the embodiments of the invention, the row is taken as a unit to carry out frame compression, so that the consumption of hardware resources is reduced. The operand for computing the bit depths is far less than the operand of an entropy encoding method, so that the coding efficiency is high.