Systems and methods for raster-to-block converter
View Patent ↗A raster to block converter and equivalently a block to raster converter can be implemented using enough memory to contain a single image band, that is a band of pixels of height equal to a single block but spanning the entire width of an image. The raster to block converter can operate at full rate so that as soon as a pixel is read out from the memory a new pixel can be stored in its place. The location of a pixel can be tracked using a mapping involving basic modular arithmetic. This raster to block converter is scalable so that it can work with any size image and block size.
1. A method of processing an image comprising:
determining a memory location from an image band location associated with a first pixel having first pixel data;
retrieving the first pixel data from a memory at the memory location;
outputting the first pixel data to an outbound stream of pixel data;
receiving second pixel data from an inbound stream;
storing the second pixel data to the memory at the memory location,
wherein the determining the memory location comprises:
separating the image band location into an image sliver part i, a pixel part, and an offset part on the basis of a predetermined block size;
mapping the image sliver part to a converted image sliver part addr(i) by the equation
addr
(
i
)
=
{
δ
i
modulo
(
w
-
1
)
w
-
1
if
i
≠
w
-
1
otherwise
,
where w is the width of the image and δ is a multiplicative factor; and
combining the converted image sliver part, the pixel part and the offset part into the memory location; and
multiplying δ by a predetermined factor modulo (w−1) after all pixels in an image band have been retrieved.
2. The method of claim 1 , wherein the inbound stream is in raster order;
the outbound stream is in block order;
the predetermined block has a width; and
the predetermined factor is w divided by the width.
3. A method of claim 2 , further comprising:
performing a discrete cosine transform on the outbound stream of pixel data to produce transformed image data;
quantizing the image transformed image data; and
entropy encoding the transformed image data.
4. The method of claim 1 , wherein the inbound stream is in block order;
the outbound stream is in raster order;
the predetermined block has a height; and
the predetermined factor is the height.
5. The method of claim 4 , further comprising:
entropy decoding encoded image data;
dequantizing the encoded image data; and
performing an inverse DCT on the encoded image data to produce the inbound stream of pixel data.
6. An image processing apparatus comprising:
a reordering converter for reordering an image from an inbound stream of pixel data into an outbound stream of pixel data comprising:
a memory;
control logic which retrieves first pixel data from a memory location in the memory and places the first pixel data on the outbound stream of pixel data and receives second pixel data from the inbound stream and stores the second pixel data into the memory at the memory location;
an address converter, which separates an image band location associated with the first pixel data into an image sliver part i, a pixel part, and an offset part on the basis of a predetermined block size; maps the image sliver part is converted by the equation
addr
(
i
)
=
{
δ
i
modulo
(
w
-
1
)
w
-
1
if
i
≠
w
-
1
otherwise
into image sliver part addr(i), where w is the width of the image and δ is a multiplicative factor; combines the converted image sliver part, the pixel part and the offset part into the memory location; and multiplies δ by a predetermined factor after all pixels in an image band have been retrieved.
7. The image processing apparatus of claim 6 , wherein the memory comprises a frame buffer.
8. The image processing apparatus of claim 6 , wherein the inbound stream is in raster order;
the outbound stream is in block order;
the predetermined block has a width; and
the predetermined factor is w divided by the width.
9. The image processing apparatus of claim 8 , further comprising:
a DCT module coupled to the reordering converter;
a quantizer coupled to the DCT module; and
an entropy coder coupled to the quantizer.
10. The image processing apparatus of claim 8 , further comprising:
a DCT module coupled to the reordering converter,
a quantizer coupled to the DCT module,
a dequantizer coupled to the quantizer,
an IDCT module coupled to the dequantizer;
a motion estimator; and
a motion compensator coupled to the motion estimator.
11. The image processing apparatus of claim 8 , further comprising:
a second reordering converter coupled to the motion estimator.
12. The image processing apparatus of claim 6 , wherein the inbound stream is in block order;
the outbound stream is in raster order;
the predetermined block has a height; and
the predetermined factor is the height.
13. The image processing apparatus of claim 12 , further comprising:
a iDCT module coupled to the reorder converter;
a dequantizer coupled to the DCT module; and
an entropy decoder coupled to the dequantizer.
14. The image processing apparatus of claim 12 , further comprising:
a iDCT module coupled to the reorder converter;
a dequantizer coupled to the DCT module;
an entropy decoder coupled to the dequantizer; and
a motion compensator coupled to the entropy decoder.
15. An image processing apparatus comprising:
a memory;
means for converting an image band location associated with a first pixel having first pixel data into a memory location, comprising:
means for separating the image band location into an image sliver part i, a pixel part, and an offset part on the basis of a predetermined block size;
means for mapping the image sliver part to a converted image sliver part addr(i) by the equation
addr
(
i
)
=
{
δ
i
modulo
(
w
-
1
)
w
-
1
if
i
≠
w
-
1
otherwise
,
where w is the width of the image and δ is a multiplicative factor;
means for combining the converted image sliver part, the pixel part and the offset part into the memory location; and
means for multiplying δ by a predetermined factor after all pixels in the first image band have been retrieved;
means for retrieving the first pixel data from the memory at the memory location;
means for outputting the first pixel data to an outbound stream of pixel data;
means for receiving second pixel data from an inbound stream;
means for storing the second pixel data to the memory at the memory location.
16. The image processing apparatus of claim 15 , wherein the inbound stream is in raster order;
the outbound stream is in block order;
the predetermined block has a width; and
the predetermined factor is w divided by the width.
17. The image processing apparatus of claim 16 , further comprising:
means for performing a discrete cosine transform on the image to produce transformed image data;
means for quantizing the image transformed image data; and
means for entropy encoding the transformed image data.
18. The image processing apparatus of claim 15 , wherein the inbound stream is in block order;
the outbound stream is in raster order;
the predetermined block has a height; and
the predetermined factor is the height.
19. The image processing apparatus of claim 18 , further comprising:
means for entropy decoding the encoded image data;
means for dequantizing the encoded image data; and
performing an inverse DCT on the encoded image data to produce an image.
20. The image processing apparatus of claim 18 , further comprising means for entropy decoding the encoded image data.