FIELD: computer engineering. SUBSTANCE: method involves normalization of complexity of corresponding large-scale units using three-dimensional complexity with respect to input image, calculation of sum of zero coefficients by performing discrete cosine-transform of input image and calculation of coefficients of discrete cosine-transform within range of -1+1, calculation of normalization value of zero coefficients for respective large-scale units according to sum of zero coefficients, calculation of transient weight function of respective large-scale units according to sum of zero coefficients and given compression rate, and adding complexity normalization value, normalization value of zero coefficients, and transient weight function in order to scaling coefficient. Sampling step in discrete cosine-transform is controlled in large-scale processing unit according to scaling coefficient. EFFECT: frames of images with alternating complexity are encoded for constant bit transmission rate without decrease in image quality. 11 cl, 8 dwg
