Image coding device and method
AssigneeMatsushita Electric Industrial Co., Ltd.
InventorHiroshi Saito
Life of this patent
Estimating the term…
Abstract
To compress and encode an image by dividing the image into units of blocks, a method in which features of each block are detected and different operations are applied according to the features has been used often so as to reduce coding distortion and improve subjective image quality, but this method requires additional information representing features of blocks to be recorded and transmitted together, thereby deteriorating the coding efficiency. A rounding operation with respect to a calculation result is modified so that the rounding operation is varied according to features detected as to each block. By so doing, adaptive compressing and coding is implemented by only the processing upon coding, without recording or transmitting additional information.
Claims18
An image coding device that encodes an image by dividing the image into units of blocks, the image coding device comprising:
- a feature detecting means that detects features of the image as to each block;
- a classification means that classifies each block into one of a plurality of categories according to a result of the detection by the feature detection means;
- a calculation means that performs a predetermined calculation and outputs a result of the calculation, the result being a real number with an integer part and a fractional part;
- and
- a rounding means that adds a rounding coefficient R (R≧0) to the result of the calculation that is the real number, and rounds an addition result to the integer part by discarding digits of the fractional part,
- wherein the rounding means varies a value of the rounding coefficient R according to the category into which the classification means classifies the block.
An image coding device that encodes an image by dividing the image into units of blocks, the image coding device comprising:
- a feature detecting means that detects features of the image as to each of the blocks;
- a classification means that classifies each block into one of a plurality of categories according to a result of the detection by the feature detection means;
- a calculation means that performs a predetermined calculation for quantization, and outputs a result of the calculation, the result being a real number with an integer part and a fractional part;
- a rounding means that adds a rounding coefficient R (R≧0) to the result of the calculation that is the real number, and rounds an addition result to the integer part by discarding digits of the fractional part;
- a codeword length estimation means that estimates a length of a variable-length code according to a frequency of occurrence of the calculation result rounded to the integer;
- and
- a variable-length coding means that actually allocates the variable-length code estimated by the codeword length estimation means to the calculation result so as to encode the calculation result,
- wherein the rounding means varies the rounding coefficient R according to the category into which the classification means classifies the block, and in the case where the codeword length estimation means estimates that a length of the variable-length code is reduced, the rounding means sets the rounding coefficient R to the value that causes the length of the variable-length code to be reduced.
An image coding method for encoding an image by dividing the image into units of blocks, the image coding method comprising:
- detecting features of the image as to each block;
- classifying the block into one of a plurality of categories according to a result of the detection;
- performing a predetermined calculation with respect to the block;
- adding a rounding coefficient R varied according to the category to a result of the calculation, the result being a real number with an integer part and a fractional part, and rounding a result of the addition to the integer part by discarding digits of the fractional part, for compression.
An image coding method for encoding an image by dividing the image into units of blocks, the image coding method comprising:
- detecting features of the image as to each block;
- classifying the block into one of a plurality of categories according to a result of the detection;
- performing a predetermined calculation with respect to the block;
- adding a rounding coefficient R varied according to the category to a result of the calculation, the result being a real number with an integer part and a fractional part, and rounding a result of the addition to the integer part by discarding digits of the fractional part, for compression;
- and
- estimating a length of a variable-length code according to the category,
- wherein in the case where the length of the variable-length code is estimated to be reduced, a value of the rounding coefficient R that causes the length of the variable-length code to be reduced is employed.
Dependencies are inferred from the claim text.