Image coding control method and device

An image coding control device for controlling a quantization parameter by using a prediction mode selected from plural prediction modes every any coding unit, comprising:

• an undetermined multiplier calculating processor unit for calculating an undetermined multiplier serving as a parameter used for a cost function from a given reference quantization parameter;
• a set selecting processor unit for selecting a set of secondary number of coded bits—coding distortion pairs from a set of primary number of coded bits—coding distortion pairs that correspond to plural combinations of the prediction mode and the quantization parameter every coding unit;
• an optimal set selecting processor unit for selecting number of coded bits—coding distortion pairs providing the minimum cost function as the optimal number of coded bits—coding distortion pairs from the set of the secondary number of coded bits—coding distortion pairs by using the undetermined multiplier thus calculated every coding unit;
• and
• a determining processor unit for determining a prediction mode and a quantization parameter used for the coding from the optimal number of coded bits—coding distortion pairs,
• wherein when the set of the secondary number of coded bits—coding distortion pairs is selected every coding unit, the set selecting processor unit
• (1) forms a convex hull comprising the set of the primary number of coded bits—coding distortion pairs on a two-dimensional orthogonal coordinate system in which the number of coded bits is set as one axis and the coding distortion is set as the other axis,
• (2) determines a set on the convex hull that connect the position of a number of coded bits—coding distortion pair providing the minimum number of coded bits and the position of a number of coded bits—coding distortion pair providing the minimum coding distortion, and
• (3) selects a set of number of coded bits—coding distortion pairs that is on the convex hull and exists at a nearer side to the origin of a two-dimensional orthogonal coordinate system, as a set of the secondary number of coded bits—coding distortion pairs.

An image coding control program product for controlling a quantization parameter through a computer by using a prediction mode selected from plural prediction modes every any coding unit, the program product comprising instructions of:

• calculating an undetermined multiplier serving as a parameter used for a cost function from a given reference quantization parameter;
• selecting a set of secondary number of coded bits—coding distortion pairs from a set of primary number of coded bits—coding distortion pairs that correspond to plural combinations of the prediction mode and the quantization parameter every coding unit;
• selecting number of coded bits—coding distortion pairs providing the minimum cost function as the optimal number of coded bits—coding distortion pairs from the set of the secondary number of coded bits—coding distortion pairs by using the undetermined multiplier thus calculated every coding unit;
• and
• determining a prediction mode and a quantization parameter used for the coding from the optimal number of coded bits—coding distortion pairs,
• wherein when the set of the secondary number of coded bits—coding distortion pairs is selected every coding unit, the instruction of selecting the set of secondary number of coded bits—coding distortion pairs
• (1) forms a convex hull comprising the set of the primary number of coded bits—coding distortion pairs on a two-dimensional orthogonal coordinate system in which the number of coded bits is set as one axis and the coding distortion is set as the other axis,
• (2) determines a set on the convex hull that connect the position of a number of coded bits—coding distortion pair providing the minimum number of coded bits and the position of a number of coded bits—coding distortion pair providing the minimum coding distortion, and
• (3) selects a set of number of coded bits—coding distortion pairs that is on the convex hull and exists at a nearer side to the origin of the two-dimensional orthogonal coordinate system, as a set of the secondary number of coded bits—coding distortion pairs.