IMAGE ENCODING/DECODING METHOD AND A DEVICE FOR MAXIMIZING A COMPRESSION RATE OF ENCODING BY USING DISCRETE SINE TRANSFORM CONVERSION OR DISCRETE COSINE TRANSFORM CONVERSION CORRESPONDING TO A CONVERSION UNIT

摘要

PURPOSE: An image encoding/decoding method and a device are provided to use other conversion method or other inverse conversion method corresponding to a conversion unit and perform inverse conversion of a frequency coefficient block or conversion of a residual block, thereby increasing a compression rate.;CONSTITUTION: An image encoding device generates a prediction value by a movement vector value for a current block in an image(S610). The image encoding device generates residual signals by difference calculation of an original pixel value and a predicted pixel value of the current block(S620). The image encoding device converts the residual signals into a frequency domain by discrete sine transform conversion or discrete cosine transform conversion(S630). The image encoding device quantizes the converted residual signals(S640). The image encoding device determines frequency conversion residual signals. The image encoding device encodes the determined residual signals as a bit stream.;COPYRIGHT KIPO 2013;[Reference numerals] (AA) Start; (BB) End; (S610) Prediction step of generating a prediction value by a movement vector value for a current block in an image; (S620) Subtraction step of generating residual signals by difference calculation of an original pixel value and a predicted pixel value of the current block; (S630) Transforming step of transforming the residual signals into a frequency domain by using DST transformation and DCT transformation when residual signals, generated by inter prediction, are transformed by 4x4 transform unit and are transformed by a transform unit of a different size, respectively; (S640) Quantization and bitstream generation step of quantizing the transformed residual signals into the frequency domain(S640) and determining the optimal quantized frequency transformation residual signals to be decoded into bitstreams