Methods and apparatus to improve detection and false alarm rate over image segmentation
Methods and apparatus to improve detection and false alarm rate over image segmentation are disclosed. An example method includes determining a first score based on a first pixel distance between a first feature of a first object in a reference image and a second feature in a second image of a search area, determining a second score corresponding to a second pixel distance between the second feature and a mathematical representation of a plurality of shapes representing a similarity between the second feature in the second image of the search area and the plurality of shapes observed simultaneously. The method further includes determining a normalized score by normalizing the first score based on the second score and identifying a second object in the second image of the search area as the first object in the reference image when the normalized score satisfies a threshold score.
1. An apparatus to detect an object comprising:
an image scorer to:
determine a first score based on a first pixel distance between a first feature of a first object in a reference image and a second feature in a second image of a search area; and
determine a second score corresponding to a second pixel distance between the second feature and a mathematical representation of a plurality of shapes representing a similarity between the second feature in the second image of the search area and the plurality of shapes observed simultaneously;
a normalizer to determine a normalized score by normalizing the first score based on the second score; and
an object detector to identify a second object in the second image of the search area as the first object in the reference image when the normalized score satisfies a threshold score.
2. The apparatus of claim 1 , wherein the second score represents a lesser similarity between the second feature and the plurality of shapes observed simultaneously than a second similarity between the second feature and any one of the plurality of shapes.
3. The apparatus of claim 1 , wherein the plurality of shapes are representative of background clutter.
4. The apparatus of claim 1 , wherein the second feature in the second image includes an edge of a retail store shelf.
5. The apparatus of claim 1 , wherein the image scorer is to determine the second score by cross-correlating the second feature with a matrix of ones having a same perimeter as the first object.
6. The apparatus of claim 1 , wherein when the normalized score does not satisfy the threshold score:
the image scorer is further to:
determine a third score based on a third pixel distance between the first feature of the first object in the reference image and a third feature in the second image of the search area, the third feature at a different location in the second image than the second feature in the second image; and
determine a fourth score corresponding to a fourth pixel distance between the third feature and a second mathematical representation of a second plurality of shapes representing a similarity between the third feature in the second image of the search area and the second plurality of shapes observed simultaneously;
the normalizer is further to determine a second normalized score by normalizing the third score based on the fourth score; and
the object detector is further to identify the second object in the second image of the search area as the first object in the reference image when the second normalized score satisfies the threshold score.
7. A method to detect an object in an image comprising:
determining, by executing an instruction with a processor, a first score based on a first pixel distance between a first feature of a first object in a reference image and a second feature in a second image of a search area;
determining, by executing an instruction with the processor, a second score corresponding to a second pixel distance between the second feature and a mathematical representation of a plurality of shapes representing a similarity between the second feature in the second image of the search area and the plurality of shapes observed simultaneously;
determining, by executing an instruction with the processor, a normalized score by normalizing the first score based on the second score; and
identifying, by executing an instruction with the processor, a second object in the second image of the search area as the first object in the reference image when the normalized score satisfies a threshold score.
8. The method of claim 7 , wherein the second score represents a lesser similarity between the second feature and the plurality of shapes observed simultaneously than a second similarity between the second feature and any one of the plurality of shapes.
9. The method of claim 7 , wherein the plurality of shapes are representative of background clutter.
10. The method of claim 7 , wherein the second feature in the second image includes an edge of a retail store shelf.
11. The method of claim 7 , wherein determining the second score further includes cross-correlating the second feature with a matrix of ones having a same perimeter as the first object.
12. The method of claim 7 , further including, when the normalized score does not satisfy the threshold score:
determining, by executing an instruction with the processor, a third score based on a third pixel distance between the first feature of the first object in the reference image and a third feature in the second image of the search area, the third feature at a different location in the second image than the second feature in the second image;
determining, by executing an instruction with the processor, a fourth score corresponding to a fourth pixel distance between the third feature and a second mathematical representation of a second plurality of shapes representing a similarity between the third feature in the second image of the search area and the second plurality of shapes observed simultaneously;
determining, by executing an instruction with the processor, a second normalized score by normalizing the third score based on the fourth score; and
identifying, by executing an instruction with the processor, the second object in the second image of the search area as the first object in the reference image when the second normalized score satisfies the threshold score.
13. A non-transitory computer readable storage medium comprising instructions, that when executed, cause a processor to at least:
determine a first score based on a first pixel distance between a first feature of a first object in a reference image and a second feature in a second image of a search area;
determine a second score corresponding to a second pixel distance between the second feature and a mathematical representation of a plurality of shapes representing a similarity between the second feature in the second image of the search area and the plurality of shapes observed simultaneously;
determine a normalized score by normalizing the first score based on the second score; and
identify a second object in the second image of the search area as the first object in the reference image when the normalized score satisfies a threshold score.
14. The non-transitory computer readable storage medium of claim 13 , wherein the second score represents a lesser similarity between the second feature and the plurality of shapes observed simultaneously than a second similarity between the second feature and any one of the plurality of shapes.
15. The non-transitory computer readable storage medium of claim 13 , wherein the plurality of shapes are representative of background clutter.
16. The non-transitory computer readable storage medium of claim 13 , wherein the second feature in the second image includes an edge of a retail store shelf.
17. The non-transitory computer readable storage medium of claim 13 , wherein the instructions are further to cause the processor to determine the second score by cross-correlating the second feature with a matrix of ones having a same perimeter as the first object.
18. The non-transitory computer readable storage medium of claim 13 , wherein when the normalized score does not satisfy the threshold score, the instructions are further to cause the processor to:
determine a third score based on a third pixel distance between the first feature of the first object in the reference image and a third feature in the second image of the search area, the third feature at a different location in the second image than the second feature in the second image;
determine a fourth score corresponding to a fourth pixel distance between the third feature and a second mathematical representation of a second plurality of shapes representing a similarity between the third feature in the second image of the search area and the second plurality of shapes observed simultaneously;
determine a second normalized score by normalizing the third score based on the fourth score; and
identify the second object in the second image of the search area as the first object in the reference image when the second normalized score satisfies the threshold score.