Image compression method allowing a set compression quality to be obtained
An image compression method comprises the following steps applied to at least one image: decorrelating the image by applying thereto a mathematical transform to obtain a set of coefficients, decomposing the image into blocks and, for each block of the image, quantizing the coefficients using a dead-zone uniform scalar quantizer having a semi-dead zone of size T and a quantization step size Δ, and coding the quantized coefficients, the size T of the semi-dead zone of the scalar quantizer being determined by minimizing the difference between an estimated quantization distortion D(T), dependent at least on size T, and a target quantization distortion D c .
1. An image compression method comprising the following steps applied to at least one image, the method being executed by an image coding device:
receiving at least one image captured by an image-capturing device,
decorrelating the image by applying thereto a mathematical transform so as to obtain a set of coefficients, and
decomposing the image into blocks and, for each block of the image,
quantizing said coefficients using a dead-zone uniform scalar quantizer having a semi-dead zone of size T and a quantization step size Δ, and
coding the quantized coefficients to produce a compressed image,
the size T of the semi-dead zone of the scalar quantizer being determined by minimizing a difference between an estimated quantization distortion D(T), dependent at least on said size T, and a target quantization distortion D c .
2. The image compression method according to claim 1 , wherein said mathematical transform is a wavelet transform or a discrete cosine transform.
3. The image compression method according to claim 1 , wherein the quantized coefficients are coded using a source coder.
4. The image compression method according to claim 1 further comprising, for each block, a step of coding the size T of the semi-dead zone losslessly.
5. The image compression method according to claim 1 , wherein the quantized coefficients are coded using a source coder that comprises an entropy coder.
6. The image compression method according to claim 1 further comprising, for each block, setting the quantization step size Δ equal to a determined size T of the semi-dead zone weighted by a preset parameter α.
7. The image compression method according to claim 6 , wherein the parameter α is chosen in an interval [0.2; 3].
8. The image compression method according to claim 6 , wherein the parameter α is set independently for each block.
9. The image compression method according to claim 6 , wherein the parameter α is set to an identical value for all the blocks of the image.
10. The image compression method according to claim 6 further comprising a step of coding the parameter α losslessly.
11. The image compression method according to claim 1 , wherein the estimated quantization distortion corresponds to a mean error made while quantizing said coefficients using the dead-zone uniform scalar quantizer.
12. The image compression method according to claim 11 , wherein the estimated quantization distortion is determined by summing a first term representative of the quantization distortion engendered by a zeroing of coefficients a modulus of which is less than the size T of the semi-dead zone and a second term representative of the quantization distortion engendered by the quantization, with the quantization step size Δ, of coefficients having a modulus greater than or equal to the size T of the semi-dead zone.
13. The image compression method according to claim 12 , wherein said second term is determined by the following calculation:
α
2
M
(
T
)
T
2
12
,
where α is a preset parameter of the quantizer and M(T) is a number of coefficients the modulus of which is greater than or equal to the size T of the semi-dead zone of the scalar quantizer.
14. An image coder for coding at least one image comprising:
an image processor configured to:
receive at least one image captured by an image-capturing device,
decorrelate the image by applying thereto a mathematical transform so as to obtain a set of coefficients, and
decompose the image into blocks and, for each block of the image,
a dead-zone uniform scalar quantizer of semi-dead zone of size T and of quantization step size Δ for quantizing said coefficients, and
a source coder for coding the quantized coefficients to produce a compressed image,
the size T of the semi-dead zone of the scalar quantizer being determined by minimizing a difference between an estimated quantization distortion D(T), dependent at least on said size T, and a target quantization distortion D c .
15. A satellite comprising an image coder according to claim 14 .
16. The satellite according to claim 6 further comprising a transmitter for transmitting to ground coded quantized coefficients and/or the size T of a coded semi-dead zone and/or a coded parameter α.
17. A computer program comprising instructions stored on a tangible non-transitory storage medium for executing, on a processor, an image compression method comprising the following steps applied to at least one image:
receiving at least one image captured by an image-capturing device,
decorrelating the image by applying thereto a mathematical transform so as to obtain a set of coefficients, and
decomposing the image into blocks and, for each block of the image,
quantizing said coefficients using a dead-zone uniform scalar quantizer having a semi-dead zone of size T and a quantization step size Δ, and
coding the quantized coefficients to produce a compressed image,
the size T of the semi-dead zone of the scalar quantizer being determined by minimizing a difference between an estimated quantization distortion D(T), dependent at least on said size T, and a target quantization distortion D c , when the program is executed by the processor.
18. A tangible non-transitory processor readable recording medium on which is stored a program comprising instructions for executing an image compression method comprising the following steps applied to at least one image:
receiving at least one image captured by an image-capturing device,
decorrelating the image by applying thereto a mathematical transform so as to obtain a set of coefficients, and
decomposing the image into blocks and, for each block of the image,
quantizing said coefficients using a dead-zone uniform scalar quantizer having a semi-dead zone of size T and a quantization step size Δ, and
coding the quantized coefficients to produce a compressed image,
the size T of the semi-dead zone of the scalar quantizer being determined by minimizing a difference between an estimated quantization distortion D(T), dependent at least on said size T, and a target quantization distortion D c , when the program is executed by a processor.