Image processing

1. An imaging method, comprising:

receiving image data of a 3-dimensional object; and

allocating a confidence level to at least a portion of an image frame of said image data using a machine-learning algorithm, said confidence level indicating a likelihood of that image frame having a specified element imaged on a specified plane through said 3-dimensional object;

wherein:

said allocating comprises allocating a plurality of confidence levels to said image frame of said image data, each confidence level indicating a likelihood of that image frame having a corresponding specified element imaged on a corresponding specified plane through said 3-dimensional object;

said allocating comprises allocating a confidence level to each of a plurality of image frames of said image data; and

said machine learning algorithm generates a feature map for each specified element, said feature map having map components;

when said confidence level exceeds a threshold amount for that image frame, generating a saliency map for that image frame which maps activated map components from said feature map to pixels in that image frame, wherein said generating said saliency map comprises mapping a selected percentage of the most activated map components from said feature map to pixels in that image frame; and

annotating image frames by highlighting its specified element using said saliency map, wherein said highlighting comprises overlaying at least one of a bounding box surrounding that specified element, a colour change of pixels within said image frame defining that specified element and a measurement of said specified element.

2. An imaging method comprising:

receiving image data of a 3-dimensional object; and

allocating a confidence level to at least a portion of an image frame of said image data using a machine-learning algorithm, said confidence level indicating a likelihood of that image frame having a specified element imaged on a specified plane through said 3-dimensional object;

wherein:

said image data comprises a 3-dimensional image data set and said method comprises generating said image frames from said 3-dimensional image data set;

said generating comprises generating image frames representing a sequence of parallel planes through said 3-dimensional image data set; and

said generating comprises generating image frames representing a first sequence of parallel planes through said 3-dimensional image data set and image frames representing at least a second sequence of parallel planes through said 3-dimensional image data set orthogonal to said first sequence of parallel planes;

further comprising combining saliency maps from each sequence and performing a 3-dimensional transform to obtain an image frame having that specified element imaged on that specified plane through said 3-dimensional object.

3. An imaging apparatus, comprising:

reception logic operable to receive image data of a 3-dimensional object; and

logic operable to allocate a confidence level to at least a portion of an image frame of said image data using a machine-learning algorithm, said confidence level indicating a likelihood of that image frame having a specified element imaged on a specified plane through said 3-dimensional object;

wherein said machine learning algorithm generates a feature map for each specified element, said feature map having map components and said logic is operable, when said confidence level exceeds said threshold amount for that image frame, to generate a saliency map for that image frame which maps activated map components from said feature map to pixels in that image frame.

4. The imaging apparatus of claim 3 , wherein said logic is operable to allocate a plurality of confidence levels to said image frame of said image data, each confidence level indicating a likelihood of that image frame having a corresponding specified element imaged on a corresponding specified plane through said 3-dimensional object.

5. The imaging apparatus of claim 4 , wherein said logic is operable to allocate a confidence level to each of a plurality of image frames of said image data.

6. The imaging apparatus of claim 3 , wherein said logic is operable to indicate one of an increase and a decrease in each confidence level to said operator.

7. The imaging apparatus of claim 3 , wherein said logic is operable to indicate to said operator a movement of said image capture apparatus required to increase a confidence level.

8. The imaging apparatus of claim 3 , wherein said logic is operable to indicate to said operator when each confidence level exceeds a threshold amount.

9. The imaging apparatus of claim 8 , wherein said logic is operable to store at least one image frame when said confidence level for that image frame exceeds said threshold amount and associate that image frame with its specified element.

10. The imaging apparatus of claim 3 , wherein said logic is operable to map a selected percentage of the most activated map components from said feature map to pixels in that image frame.

11. The imaging apparatus of claim 3 , wherein said logic is operable to annotate image frames by highlighting its specified element using said saliency map.

12. The imaging apparatus of claim 11 , wherein said logic is operable to highlight its specified element by overlaying at least one of a bounding box surrounding that specified element, a colour change of pixels within said image frame defining that specified element and a measurement of said specified element.

13. The imaging apparatus of claim 3 , wherein said logic is operable to train said machine learning algorithm using batches of source training data, said source training data comprising image frames for each specified element imaged on said specified plane through said 3-dimensional object, one specified element being a background element where no item of interest is present.

14. The imaging apparatus of claim 13 , wherein a format of image frames for said background element differ from a format of image frames for other specified elements and said logic is operable to adjust a format of said image frames for other specified elements to align with that of said background element.

15. The imaging apparatus of claim 13 , wherein said logic is operable to include as said image frames for said background element, image frames from said source training data where greater than a threshold amount of movement occurs between adjacent image frames of said source training data.

16. An imaging apparatus comprising:

reception logic operable to receive image data of a 3-dimensional object; and

logic operable to allocate a confidence level to at least a portion of an image frame of said image data using a machine-learning algorithm, said confidence level indicating a likelihood of that image frame having a specified element imaged on a specified plane through said 3-dimensional object;

wherein:

said image data comprises a 3-dimensional image data set and said logic is operable to generate said image frames from said 3-dimensional image data set;

said logic is operable to generate image frames representing a sequence of parallel planes through said 3-dimensional image data set;

said logic is operable to generate image frames representing a first sequence of parallel planes through said 3-dimensional image data set and image frames representing at least a second sequence of parallel planes through said 3-dimensional image data set orthogonal to said first sequence of parallel planes; and

said logic is operable to combine saliency maps from each sequence and to perform a 3-dimensional transform to obtain an image frame having that specified element imaged on that specified plane through said 3-dimensional object.