Optimization for deconvolution
Disclosed herein includes a system, a method, and a device for improving computational efficiency of deconvolution by reducing a number of dot products. In one aspect, an input image having a set of pixels is received. A first dot product may be performed on a subset of the set of pixels of the input image and a portion of a kernel, to generate a first pixel of an output image. A number of multiplications performed for the first dot product performed may be less than a number of elements of the kernel. A second dot product on a remaining portion of the kernel to generate the first pixel of the output image may be bypassed.
1. A method comprising:
performing, by one or more processors according to predetermined mapping information, a first dot product on a first subset of a set of pixels of an input image and a portion of a kernel, to generate a first pixel of an output image;
bypassing, by the one or more processors according to the predetermined mapping information, a second dot product on a second subset of the set of pixels of the input image and a remaining portion of the kernel to generate the first pixel of the output image; and
performing, by the one or more processors, a third dot product on i) the first subset of the set of pixels of the input image and ii) another portion of the kernel to generate a second pixel of the output image.
2. The method of claim 1 , further comprising:
identifying, by the one or more processors according to the predetermined mapping information indicating a corresponding dot product to bypass for each pixel of the output image, the second dot product to bypass for the first pixel of the output image; and
bypassing, by the one or more processors according to the predetermined mapping information, the second dot product on the remaining portion of the kernel to generate the first pixel of the output image.
3. The method of claim 1 , further comprising:
identifying, by the one or more processors according to the predetermined mapping information indicating a corresponding dot product to perform for each pixel of the output image, the first dot product to perform for the first pixel for the output image; and
performing, by the one or more processors according to the predetermined mapping information, the first dot product on the first subset of the set of pixels of the input image and the portion of the kernel, to generate the first pixel of the output image.
4. The method of claim 1 , wherein the output image has more number of pixels than the input image.
5. The method of claim 1 , wherein the number of multiplications performed for the first dot product to generate the first pixel of the output image is same as a number of multiplications performed for the third dot product to generate the second pixel of the output image.
6. The method of claim 1 , further comprising:
performing, by the one or more processors, a third dot product on i) another subset of the set of pixels of the input image and ii) the portion of the kernel, to generate a second pixel of the output image.
7. The method of claim 6 , wherein the number of multiplications performed for the first dot product to generate the first pixel of the output image is same as a number of multiplications performed for the third dot product to generate the second pixel of the output image.
8. The method of claim 1 , further comprising:
performing, by the one or more processors, a third dot product on i) another subset of the set of pixels of the input image and ii) another portion of the kernel, to generate a second pixel of the output image.
9. The method of claim 8 , wherein the number of multiplications performed for of the first dot product to generate the first pixel of the output image is different from a number of multiplications performed for the third dot product to generate the second pixel of the output image.
10. A system comprising:
one or more processors configured to:
perform, according to predetermined mapping information, a first dot product on a first subset of a set of pixels of an input image and a portion of a kernel, to generate a first pixel of an output image; and
bypass, according to the predetermined mapping information, a second dot product on a second subset of the set of pixels of the input image and a remaining portion of the kernel to generate the first pixel of the output image; and
perform a third dot product on i) the first subset of the set of pixels of the input image and ii) another portion of the kernel to generate a second pixel of the output image.
11. The system of claim 10 , wherein the one or more processors are configured to:
identify, according to the predetermined mapping information indicating a corresponding dot product to bypass for each pixel of the output image, the second dot product to bypass for the first pixel of the output image; and
bypass, according to the predetermined mapping information, the second dot product on the remaining portion of the kernel to generate the first pixel of the output image.
12. The system of claim 10 , wherein the one or more processors are configured to:
identify, according to the predetermined mapping information indicating a corresponding dot product to perform for each pixel of the output image, the first dot product to perform for the first pixel of the output image; and
perform, according to the predetermined mapping information, the first dot product on the first subset of the set of pixels of the input image and the portion of the kernel, to generate the first pixel of the output image.
13. The system of claim 10 , wherein the output image has more number of pixels than the input image.
14. The system of claim 10 , wherein the number of multiplications performed for the first dot product to generate the first pixel of the output image is same as a number of multiplications performed for the third dot product to generate the second pixel of the output image.
15. The system of claim 10 , wherein the one or more processors are configured to:
perform a third dot product on i) another subset of the set of pixels of the input image and ii) the portion of the kernel, to generate a second pixel of the output image.
16. The system of claim 15 , wherein the number of multiplications performed for the first dot product to generate the first pixel of the output image is same as a number of multiplications performed for the third dot product to generate the second pixel of the output image.
17. The system of claim 10 , wherein the one or more processors are configured to:
perform a third dot product on i) another subset of the set of pixels of the input image and ii) another portion of the kernel, to generate a second pixel of the output image.
18. The system of claim 17 , wherein the number of multiplications performed for the first dot product to generate the first pixel of the output image is different from a number of multiplications performed for the third dot product to generate the second pixel of the output image.