Image processing apparatus and image processing method

An image processing method comprising:

• orthogonally transforming a prediction residual of an image and controlling a number of significant figures of coefficient data obtained by performance of orthogonal transform on the prediction residual, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of the image;
• quantizing the coefficient data and controlling a number of significant figures of a quantized coefficient obtained by quantization of the coefficient data, on a basis of the bit depth of the local level;
• deriving extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence of the image, on the basis of the bit depth of the local level;
• and
• encoding the quantized coefficient and generating a bit stream.

An image processing apparatus comprising:

• an orthogonal transform section configured to orthogonally transform a prediction residual of an image and that controls a number of significant figures of coefficient data obtained by performance of orthogonal transform on the prediction residual, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of a predicted image corresponding to the image or of a decoded image referred to at a time of generation of the predicted image;
• a quantization section configured to quantize the coefficient data obtained by the orthogonal transform section and control a number of significant figures of a quantized coefficient obtained by quantization of the coefficient data, on a basis of the bit depth of the local level of the predicted image or the decoded image;
• an extension bit precision derivation section configured to derive extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence of the image, on the basis of the bit depth of the local level;
• and
• an encoding section configured to encode the quantized coefficient obtained by the quantization section and generate a bit stream, wherein the orthogonal transform section, the quantization section, the extension bit precision derivation section, and the encoding section are each implemented via at least one processor.

An image processing method comprising:

• orthogonally transforming a prediction residual of an image and controlling a number of significant figures of coefficient data obtained by performance of orthogonal transform on the prediction residual, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of a predicted image corresponding to the image or of a decoded image referred to at a time of generation of the predicted image;
• quantizing the coefficient data and controlling a number of significant figures of a quantized coefficient obtained by quantization of the coefficient data, on a basis of the bit depth of the local level of the predicted image or the decoded image;
• deriving extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence of the image, on the basis of the bit depth of the local level;
• and
• encoding the quantized coefficient and generating a bit stream.

An image processing apparatus comprising:

• a decoding section configured to decode a bit stream;
• an inverse quantization section configured to inversely quantize a quantized coefficient obtained by decoding of the bit stream by the decoding section and control a number of significant figures of coefficient data obtained by inverse quantization of the quantized coefficient, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of an image;
• and
• an inverse orthogonal transform section configured to perform inverse orthogonal transform on the coefficient data obtained by inverse quantization of the quantized coefficient by the inverse quantization section and control a number of significant figures of residual data obtained by performance of the inverse orthogonal transform on the coefficient data, on a basis of the bit depth of the local level, wherein
• the bit stream contains information associated with extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence, and
• the decoding section, the inverse quantization section, and the inverse orthogonal transform section are each implemented via at least one processor.

An image processing method comprising:

• decoding a bit stream;
• inversely quantizing a quantized coefficient obtained by decoding of the bit stream, and controlling a number of significant figures of coefficient data obtained by inverse quantization of the quantized coefficient, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level;
• and
• performing inverse orthogonal transform on the coefficient data obtained by inverse quantization of the quantized coefficient, and controlling a number of significant figures of residual data obtained by performance of the inverse orthogonal transform on the coefficient data, on a basis of the bit depth of the local level, wherein
• the bit stream contains information associated with extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence.

An image processing apparatus comprising:

• a decoding section configured to decode a bit stream;
• an inverse quantization section configured to inversely quantize a quantized coefficient obtained by decoding of the bit stream by the decoding section and control a number of significant figures of coefficient data obtained by inverse quantization of the quantized coefficient, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of a predicted image or of a decoded image referred to at a time of generation of the predicted image;
• and
• an inverse orthogonal transform section configured to perform inverse orthogonal transform on the coefficient data obtained by inverse quantization of the quantized coefficient by the inverse quantization section and control a number of significant figures of residual data obtained by performance of the inverse orthogonal transform on the coefficient data, on a basis of the bit depth of the local level of the predicted image or the decoded image, wherein
• the bit stream contains information associated with extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence, and
• the decoding section, the inverse quantization section, and the inverse orthogonal transform section are each implemented via at least one processor.

An image processing method comprising:

• decoding a bit stream;
• inversely quantizing a quantized coefficient obtained by decoding of the bit stream, and controlling a number of significant figures of coefficient data obtained by inverse quantization of the quantized coefficient, on a basis of a bit depth indicating a range of pixel values of a local level that is a data unit smaller than a sequence level of a predicted image or of a decoded image referred to at a time of generation of the predicted image;
• and
• performing inverse orthogonal transform on the coefficient data obtained by inverse quantization of the quantized coefficient, and controlling a number of significant figures of residual data obtained by performance of the inverse orthogonal transform on the coefficient data, on a basis of the bit depth of the local level of the predicted image or the decoded image, wherein
• the bit stream contains information associated with extension bit precision that includes a correction value of a controlled variable of the number of significant figures of the coefficient data set per sequence.