Method and system for processing signals via perceptive vectorial quantization, computer program product therefor

A process, comprising:

• converting digital video signals having images organized in blocks of pixels between a first format and a second format, said second format being a format compressed via vector quantization, said converting including:
• obtaining said vector quantization from repeated application of a scalar quantizer to the pixels of said blocks with a quantization step determined in an adaptive way according to characteristics of the pixels, wherein said obtaining said vector quantization includes:
• identifying a sharpness value of edges in each of said blocks of pixels, and quantizing said sharpness value to divide the edges into a number of classes;
• and
• attributing a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels, wherein said vector quantization is obtained with application of two scalar quantizers with quantization steps scaled by constants ⅔ and sin (π/3).

A system for converting digital video signals having images organized in blocks of pixels between a first format and a second format, said second format being a format compressed via vector quantization, the system comprising:

• at least one converter chosen between an encoder and a decoder and wherein said converter is configured for a vector quantization resulting from repeated application of a scalar quantizer to the pixels of said blocks with a quantization step determined in an adaptive way according to characteristics of the pixels, wherein said converter is configured to:
• identify a sharpness value of edges in each of said blocks of pixels, and quantize said sharpness value to divide the edges into a number of classes;
• and
• attribute a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels,
• wherein said quantization step is determined according to at least one law chosen from the following:
• m*EQ+q;
• and
• m*(t^EQ)+q;
• where m and g are constants, determined selectively, and EQ is an index which identifies said edge classes, and t is a scalar value.

A system for converting digital video signals having images organized in blocks of pixels between a first format and a second format, said second format being a format compressed via vector quantization, the system comprising:

• at least one converter chosen between an encoder and a decoder and wherein said converter is configured for a vector quantization resulting from repeated application of a scalar quantizer to the pixels of said blocks with a quantization step determined in an adaptive way according to characteristics of the pixels, wherein said converter is configured to:
• identify a sharpness value of edges in each of said blocks of pixels, and quantize said sharpness value to divide the edges into a number of classes;
• and
• attribute a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels, wherein said converter is configured to obtain an application of two scalar quantizers with quantization steps scaled by constants ⅔ and sin(π/3).

A system for converting digital video signals having images organized in blocks of pixels between a first format and a second format, said second format being a format compressed via vector quantization, the system comprising:

• at least one converter chosen between an encoder and a decoder and wherein said converter is configured for a vector quantization resulting from repeated application of a scalar quantizer to the pixels of said blocks with a quantization step determined in an adaptive way according to characteristics of the pixels, wherein said converter is configured to:
• identify a sharpness value of edges in each of said blocks of pixels, and quantize said sharpness value to divide the edges into a number of classes;
• and
• attribute a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels, wherein said video signals in said first format are signals generated according to a so-called Bayer pattern, which can be ordered in blocks of size 4×2 having the following pattern of chromatic components:
• row 1=G1R1G2R2
• row 2=B1G3B2G4
• and wherein said vector quantization is a two-dimensional vector quantization applied to pairs <R1, R2>, <B1, B2> together with <G1, G2>, <G3, G4> or <G1, G3>, <G2, G4>.

An article of manufacture, comprising:

• a computer-readable memory medium encoded with a computer program representing instructions to cause a processor, when the computer program is run, to:
• convert a digital video signal having images organized in blocks of pixels from a first format to a second format, including instructions to use vector quantization to compress the first format into the second format, including instructions to cause said processor to obtain the vector quantization by repeatedly applying a scalar quantizer to the pixels of the blocks with an adaptive quantization step based on characteristics of the pixels, wherein the instructions to obtain the vector quantization include instructions to:
• identify a sharpness value of edges in each of said blocks of pixels, and quantize said sharpness value to divide the edges into a number of classes;
• and
• attribute a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels,
• wherein said video signals in said first format are signals generated according to a so-called Bayer pattern, which can be ordered in blocks of size 4×2 having the following pattern of chromatic components:
• row 1=G1R1G2R2,
• row 2=B1G3B2G4;
• and wherein said vector quantization is a two-dimensional vector quantization applied to pairs <R1R2>, <B1, B2> together with <G1, G2>, <G3, G4> or <G1, G3>, <G2, G4>.

A system, comprising:

• a means for receiving a digital video signal organized into blocks of pixels at a first format and for outputting the digital video signal at a second format;
• a means for encoding the digital video signal from the first format to the second format, said means for encoding including a means for performing vector quantization to compress the digital video signal at the first format into the second format using repeated application of a scalar quantizer to the pixels of the blocks with a quantization step adaptively determined based on characteristics of the pixels, the means for performing vector quantization including:
• means for identifying a sharpness value of edges in each of said blocks of pixels, and quantizing said sharpness value to divide the edges into a number of classes;
• and
• means for attributing a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels,
• wherein said quantization step is determined according to at least one law chosen from the following:
• m*EQ+q;
• and
• m*(t^EQ)+q;
• where m and q are constants, determined selectively, and EQ is an index which identifies said edge classes, and t is a scalar value.

A process, comprising:

• converting digital video signals having images organized in blocks of pixels between a first format and a second format, said second format being a format compressed via vector quantization, said converting including:
• obtaining said vector quantization from repeated application of a scalar quantizer to the pixels of said blocks with a quantization step determined in an adaptive way according to characteristics of the pixels, wherein said obtaining said vector quantization includes:
• identifying a sharpness value of edges in each of said blocks of pixels, and quantizing said sharpness value to divide the edges into a number of classes;
• and
• attributing a value to said quantization step differentiated according to the classes, wherein the value of the quantization step is increased near the edges of said images having said pixels,
• wherein said quantization step is determined according to at least one law chosen from the following:
• m*EQ+q;
• and
• m*(t^EQ)+q;
• where m and q are constants, determined selectively, and EQ is an index which identifies said edge classes, and t is a scalar value.