Automatic background recognition and removal (ABRR) in projection digital radiographic images (PDRI)

• US 5,903,660 A
• Filed: 07/16/1997
• Issued: 05/11/1999
• Est. Priority Date: 07/16/1997
• Status: Expired due to Fees

• Current Assignee: California Institute of Technology
• Original Assignee: Regents of the University of California (University of California)
• Inventors: Zhang, Jianguo, Huang, H. K.
• Primary Examiner(s): ROGERS, SCOTT A
• Application Number:
  US08/895,333
• Time in Patent Office: 664 Days
• Field of Search: 382/261, 382/260, 382/263, 382/266, 382/269, 382/209, 382/210, 382/217, 382/218, 382/128, 382/132, 382/275
• US Class Current: 382/132
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation

Abstract

A system for removal of background from projected digital radiograph images utilizes statistical methods to identify background pixel candidates and probable background edges of a PDRI. Thresholds for evaluating the probable background edges are determined utilizing statistical results and a reliable estimation of background removal is performed.

Claims

1. An automatic background recognition and removal process for removing background signals, caused by collimation during projection, from projection digital radiographic images (PDRI), with a PDRI comprising a set of image pixels, each encoding an image value and identified by a set of (x,y) coordinates, said method comprising the steps of:

scanning a set of lines of said image to identify an edge candidate pixel in each line scanned having an image value which varies in value from the image value of a neighbor pixel by at least a first threshold value;

collecting said candidate pixels into a first set of edge candidate pixels;

calculating a background score value for each edge candidate pixel in said first set, with a higher background score indicating a higher probability that a pixel is located in the background;

filtering said first set to discard all edge candidate pixels having a calculated background score value less than a probability threshold value from said first set to form a second edge set of edge candidate pixels;

determining which edge candidate pixels in said second set of edge candidate pixels are not located on an edge of two image areas of different intensity;

filtering said second set to discard all edge candidate pixels in the second set which are not on an image edge to form a third set of edge candidate pixels;

adaptively determining an edge probability threshold value for the number of pixels in said third edge pixel set based on a calculated value of background score of a side of the image nearest to the edge, with said threshold value increased if the calculated value is small and decreased if the calculated value is large;

fitting a line to the pixels in said third edge pixel set if the number of pixels in said set is greater than the edge probability threshold value adaptively determined for the third set; and

performing background removal of all pixels outside the image edge determined by the line.

2. The process of claim 1, wherein said step of determining which edge candidate pixels in said second set are not located on an image edge further comprises the steps of:

determining an image gradient direction (angle) value of each edge candidate pixel in said second set;

determining a set of maximum image gradient value peaks based on a distribution of image gradient direction values of edge candidate pixels in said second set, with each maximum image gradient peak characterized by a half width; and

identifying all edge candidate pixels not having image gradient values within the half width of a maximum image gradient value peak as not being located on an image edge.

3. The process of claim 1, further comprising the steps of:

performing a background removal reliability check prior to said step of performing background removal; and

canceling said step of background removal if said background removal reliability check indicates that background removal would adversely affect the PDRI.

4. The process of claim 3, wherein said step of performing a background removal reliability check comprises the steps of:

forming an first intensity level histogram of said PDRI prior to background removal;

forming a second intensity level histogram of said PDRI if background removal were performed; and

comparing sections of said first and second histograms for intensity levels below a background threshold level to determine if a significant portion of the image would be truncated if background removal were performed.

5. The process of claim 1, wherein said step of adaptively determining an edge probability threshold value comprises the steps of:

empirically determining a maximum value smaller than the average value of peaks caused by background edges in a statistically significant set of PDRIs;

empirically determining a minimum value larger than peaks due to image edges not in the background and about equal to the minimum value of background edges in said statistically significant set of PDRIs;

setting said edge probability threshold value near said maximum value if the background probability of the image area nearest to the edge is small; and

setting said edge probability threshold value near said minimum value if the background probability of the image area nearest to the edge is large.

6. The process of claim 5, wherein the edge probability threshold function is determined by:

space="preserve" listing-type="equation">B+C(1-f(x))

where;

B+C is equal to said maximum value;

B is equal to said minimum value;

x is the background score of the image area nearest to the edge; and

f(x) increases when x increases to a maximum value of 1.

7. An automatic background recognition and removal process for removing background signals, caused by collimation during projection, from a projection digital images, with a projection digital image comprising a set of image pixels, each encoding an image value and identified by a set of (x,y) coordinates, said method comprising the steps of:

scanning a set of lines of said image to identify an edge candidate pixel in each line scanned having an image value which varies in value from the image value of a neighbor pixel by at least a first threshold value;

collecting said candidate pixels into a first set of edge candidate pixels;

assigning a background priority score value to each edge candidate pixel;

filtering said first set to discard all edge candidate pixels having a background priority score value less than a probability threshold value from said first set to form a second edge set of edge candidate pixels;

determining which edge candidate pixels in said second set of edge candidate pixels are not located on an edge of two image areas of different intensity; and

filtering said second set to discard all edge candidate pixels in the second set which are not on an image edge to form a third set of edge candidate pixels.

8. The process of claim 7 further comprising the steps of:

adaptively determining an edge probability threshold value for the number of pixels in said third edge pixel set based on a calculated value of background score of a side of the image nearest to the edge, with said threshold value increased if the calculated value is small and decreased if the calculated value is large.

9. A computer program product for use in a computer, the computer program product comprising:

a computer usable medium having computer readable program code physically embodied therein for causing the computer to perform an automatic background recognition and removal process for removing background signals, caused by collimation during projection, from a projection digital radiographic images (PDRI), with a PDRI comprising a set of image pixels, each encoding an image value and identified by a set of (x,y) coordinates, said computer program product further comprising;

computer readable program code for causing said computer to scan a set of lines of said image to identify an edge candidate pixel in each line scanned having an image value which varies in value from the image value of a neighbor pixel by at least a first threshold value;

computer readable program code for causing said computer to collect said candidate pixels into a first set of edge candidate pixels;

computer readable program code for causing said computer to calculate a background score value for each edge candidate pixel in said first set, with a higher background score indicating a higher probability that a pixel is located in the background;

computer readable program code for causing said computer to filter said first set to discard all edge candidate pixels having a calculated background score value less than a probability threshold value from said first set to form a second edge set of edge candidate pixels;

computer readable program code for causing said computer to determine which edge candidate pixels in said second set of edge candidate pixels are not located on an edge of two image areas of different intensity;

computer readable program code for causing said computer to filter said second set to discard all edge candidate pixels in the second set which are not on an image edge to form a third set of edge candidate pixels;

computer readable program code for causing said computer to adaptively determine an edge probability threshold value for the number of pixels in said third edge pixel set based on a calculated value of background score of a side of the image nearest to the edge, with said threshold value increased if the calculated value is small and decreased if the calculated value is large;

computer readable program code for causing said computer to fit a line to the pixels in said third edge pixel set if the number of pixels in said set is greater than the edge probability threshold value adaptively determined for the third set; and

computer readable program code for causing said computer to perform background removal of all pixels outside the image edge determined by the line.