Image quality by two pass temporal noise reduction

1. An image processor, comprising:

an input interface for receiving an input sequence of frames acquired by an imaging apparatus;

at least one recursive temporal noise reduction filter configured to: bi-directionally process the input sequence of frames into at least two noise processed sequences, including a first noise processed sequence and a second noise processed sequence; and

a merger circuit configured to merge the at least two noise processed sequences into an output sequence, including to replace, at least one frame of the first noise processed sequence with a corresponding at least one frame of the second noise processed sequence and to combine remaining frames by weighted averaging from the at least two noise processed sequences with weights changing as a function of index position of the output sequence.

2. The image processor of claim 1 , wherein the merger circuit is configured to replace at least one frame from an initial section of the first noise processed sequence by a corresponding at least one frame from the second noise processed sequence.

3. The image processor of claim 1 , wherein a noise level is essentially constant throughout at least an initial or terminal section of the output sequence.

4. A method of image processing, the method comprising:

receiving an input sequence of frames acquired by an imaging apparatus;

applying recursive bidirectional temporal noise reduction processing to the input sequence of frames to produce at least two noise processed sequences, including a first noise processed sequence and a second noise processed sequence; and

merging said at least two noise processed sequences into an output sequence, including replacing, at least one frame of the first noise processed sequence with a corresponding at least one frame of the second noise processed sequence and combining remaining frames by weighted averaging from the at least two noise processed sequences with weights changing as a function of index position of the output sequence.

5. The method of claim 4 , wherein said merging includes replacing at least one frame from an initial section of the first noise processed sequence by a corresponding at least one frame from the second noise processed sequence.

6. The method of claim 4 , wherein a noise power level is essentially constant throughout at least an initial or terminal section of the output sequence.

7. The method of claim 4 further comprising receiving, at an input interface, the input sequence of frames acquired by the imaging apparatus, wherein the input sequence of frames is any one of i) X-ray frames, ii) ultrasound frames and iii) magnetic resonance frames.

8. The method of claim 4 , further comprising: receiving the input sequence of frames; and encoding a moving object based on the received input sequence of frames.

9. A tangible, non-transitory computer-readable medium that stores instructions, which when executed by a processor, causes the processor to perform:

receiving an input sequence of frames acquired by an imaging apparatus;

applying recursive bidirectional temporal noise reduction processing to the input sequence of frames to produce at least two noise processed sequences, including a first noise processed sequence and a second noise processed sequence; and

merging said at least two noise processed sequences into an output sequence, including replacing, at least one frame of the first noise processed sequence with a corresponding at least one frame of the second noise processed sequence and combining remaining frames by weighted averaging from the at least two noise processed sequences with weights changing as a function of index position of the output sequence.

10. A system for image processing, comprising:

a processor;

a tangible non-transitory computer-readable medium that stores instructions, which when executed by the processor, causes the processor to:

receive an input sequence of frames acquired by an imaging apparatus;

apply recursive bidirectional temporal noise reduction processing to the input sequence of frames to produce at least two noise processed sequences, including a first noise processed sequence and a second noise processed sequence; and

merge the at least two noise processed sequences into an output sequence, including replace, at least one frame of the first noise processed sequence with a corresponding at least one frame of the second noise processed sequences and combine remaining frames by weighted averaging from the at least two noise processed sequences with weights changing as a function of index position of the output sequence.

11. The system of claim 10 , wherein a noise power level is essentially constant throughout at least an initial or terminal section of the output sequence.

12. The system of claim 10 , wherein the tangible non-transitory computer-readable medium stores further instructions, which when executed by the processor, cause the processor to receive, at an input interface, the input sequence of frames acquired by the imaging apparatus, wherein the input sequence of frames is any one of i) X-ray frames, ii) ultrasound frames and iii) magnetic resonance frames.

13. The system of claim 10 , wherein the tangible non-transitory computer-readable medium stores further instructions, which when executed by the processor causes the processor to receive the input sequence of frames, and encode a moving object.

14. An image processor, comprising:

an input interface for receiving an input sequence of frames acquired by an imaging apparatus;

at least one recursive temporal noise reduction filter configured to: bi-directionally process the input sequence of frames into at least two noise processed sequences, including a first noise processed sequence and a second noise processed sequence, and to process pixels in the input sequence of frames spatially within each frame during temporal noise reduction processing; and

a merger circuit configured to merge the at least two noise processed sequences into an output sequence, including to replace, at least one frame of the first noise processed sequence with a corresponding at least one frame of the second noise processed sequence and to combine remaining frames by weighted averaging from the at least two noise processed sequences with weights changing as a function of index position of the output sequence.

15. The image processor of claim 14 , wherein the merger circuit is configured to replace at least one frame from an initial section of the first noise processed sequence by a corresponding frame from the second noise processed sequence.

16. The image processor of claim 14 , further comprising a motion compensator circuit implemented by normalized weights.