Post-processing a multi-spectral image for enhanced object identification
View Patent ↗What is disclosed is a system and method for post-processing a multi-spectral image which has already been processed for pixel classification. A binary image is received which contains pixels that have been classified using a pixel classification method. Each pixel in the image has an associated intensity value and has a pixel value of 1 or 0 depending on whether the pixel has been classified as a material of interest or not. A block of size m×n is defined. Pixel values in a block are changed according to a threshold-based filtering criteria such that pixels in the same block all have the same binary value. The block is then shifted by k pixels and pixel processing repeats until all pixels have been processed. Once all blocks have been processed, contiguous pixels having the same binary value are grouped to form objects. In such a manner, pixel classification errors are reduced.
1. A method for post-processing a pixel-classified binary image for determination of a number of objects in the image comprising a material of interest, the method comprising:
receiving a binary image containing pixels which have been classified using a pixel classification method, each pixel in said image has an intensity value, each of said pixels has a binary value depending on whether said pixel was classified as a material of interest or not;
defining a block of pixels of size m×n, where any of: m≧2 and n≧2;
(A) calculating a density value using intensity values collected for pixels in a block of pixels of said size, said block encompassing a current pixel;
(B) comparing said density to a pre-determined threshold value; and
(C) in response to said density being less than said threshold, setting all pixels in this block to zero, otherwise setting all pixels in this block to a non-zero value; and
shifting said block by k pixels, where k is less than m or n;
repeating steps (A)-(C) until all pixels in said image have been processed;
grouping together connected pixels having a same binary value to form objects; and
counting a number of said formed objects in said image.
2. The method of claim 1 , wherein said pixel classification method comprises any of: a correlation method, a ratio method, and a best fitting reflectance method.
3. The method of claim 1 , wherein calculating said density comprises determining a mean of pixel intensity values.
4. The method of claim 1 , further comprising combining pixel intensity values to generate a new intensity value for at least one pixel in a given block.
5. The method of claim 1 , wherein pixels of said image have been classified as to a material of interest being human tissue, and said formed objects are human occupants in a motor vehicle.
6. The method of claim 1 , wherein a size of said block is selected based upon an estimated size of objects likely to be identified as a material of interest within a field of view of an imaging system used to capture said image.
7. The method of claim 1 , wherein said suitable direction is given by any of: along the x-axis, the y-axis, a diagonal, and a random direction.
8. A system for post-processing a pixel-classified binary image for determination of a number of objects in the image comprising a material of interest, the system comprising:
a memory and a storage device; and
a processor in communication with said memory and storage device, said processor executing machine readable instructions for performing:
receiving a binary image containing pixels which have been classified using a pixel classification method, each pixel in said image has an intensity value, each of said pixels has a binary value depending on whether said pixel was classified as a material of interest or not;
defining a block of pixels of size m×n, where either: m≧2 or n≧2;
(A) calculating a density value using intensity values collected for pixels in a block of pixels of said size, said block encompassing a current pixel;
(B) comparing said density to a pre-determined threshold value; and
(C) in response to said density being less than said threshold, setting all pixels in this block to zero, otherwise setting all pixels in this block to a non-zero value; and
shifting said block by k pixels, where k is less than m or n;
repeating steps (A)-(C) until all pixels in said image have been processed;
grouping together connected pixels having a same binary value to form objects; and
counting a number of said formed objects in said image.
9. The system of claim 8 , wherein said pixel classification method comprises any of: a correlation method, a ratio method, and a best fitting reflectance method.
10. The system of claim 8 , wherein calculating said density comprises determining a mean of pixel intensity values.
11. The system of claim 8 , further comprising combining pixel intensity values to generate a new intensity value for at least one pixel in a given block.
12. The system of claim 8 , wherein pixels of said image have been classified as to a material of interest being human tissue, and said formed objects are human occupants in a motor vehicle.
13. The system of claim 8 , wherein a size of said block is selected based upon an estimated size of objects likely to be identified as a material of interest within a field of view of an imaging system used to capture said image.
14. The system of claim 8 , wherein said suitable direction is given by any of: along the x-axis, the y-axis, a diagonal, and a random direction.
15. A computer implemented method for post-processing a pixel-classified binary image for determination of a number of objects in the image comprising a material of interest, the method comprising:
receiving a binary image containing pixels which have been classified using a pixel classification method, each pixel in said image has an intensity value, each of said pixels has a binary value depending on whether said pixel was classified as a material of interest or not;
defining a block of pixels of size m×n, where either: m≧2 or n≧2;
(A) calculating a density value using intensity values collected for pixels in a block of pixels of said size, said block encompassing a current pixel;
(B) comparing said density to a pre-determined threshold value; and
(C) in response to said density being less than said threshold, setting all pixels in this block to zero, otherwise setting all pixels in this block to a non-zero value; and
shifting said block by k pixels, where k is less than m or n;
repeating steps (A)-(C) until all pixels in said image have been processed;
grouping together connected pixels having a same binary value to form objects; and
counting a number of said formed objects in said image.
16. The computer implemented method of claim 15 , wherein said pixel classification method comprises any of: a correlation method, a ratio method, and a best fitting reflectance method.
17. The computer implemented method of claim 15 , wherein calculating said density comprises determining a mean of pixel intensity values.
18. The computer implemented method of claim 15 , further comprising combining pixel intensity values to generate a new intensity value for at least one pixel in a given block.
19. The computer implemented method of claim 15 , wherein pixels of said image have been classified as to a material of interest being human tissue, and said formed objects are human occupants in a motor vehicle.
20. The computer implemented method of claim 15 , wherein a size of said block is selected based upon an estimated size of objects likely to be identified as a material of interest within a field of view of an imaging system used to capture said image.
21. The computer implemented method of claim 15 , wherein said suitable direction is given by any of: along the x-axis, the y-axis, a diagonal, and a random direction.