Currently, cameras operating at a resolution of 3840x2160 (4K) are increasingly being used in broadcasting and applied television. These cameras provide high image clarity and distinctness in various systems of applied and broadcast television. However, images with such resolution contain approximately 8.3 megapixels, which form a data volume in a frame of about 25 MB. This causes a number of problems, both for real-time encoding and for the deterioration of visual quality of such images due to losses in video codec quantizers, which requires a limited frequency resource of the radio communication channels used. Therefore, the development of more efficient video compression methods that provide better image quality in real time with a lower bitrate is of great scientific and practical importance. One method for improving video compression efficiency can be the use of preliminary reduction of the pixel number in the original images by 4 times before video encoding and restoring their number after decoding on the receiving side using bidirectional image resizing methods [1, 2]. The article evaluates the quality of transformed images after their bidirectional resizing using adaptive scaling methods. The use of an adaptive image scaler at the video encoder input, which reduces the volume of video data by 4 times, allows for a reduction in the computational power requirements of microprocessor devices, lowers the cost of encoders, and also provides the possibility of improving the visual image quality by reducing the loss of useful information in encoder quantizers. The study examines the performance of adaptive image resize methods S-Spline, S-Spline XL, and S-Spline Max in their bidirectional application mode and evaluates the visual quality of restored images after their bidirectional resizing.