Efficient fixed-point real-time thresholding for signal processing
AssigneeFreescale Semiconductor, Inc.
InventorsDipesh Koirala, Christopher P. Thron
Life of this patent
Estimating the term…
Abstract
A method for efficiently calculating signal thresholds for use in signal processing is described. The method computes and stores a cumulative histogram and a weighted cumulative histogram. The method then provides a first estimate for a threshold based on a single ratio. The method next performs an iterative computation to get to the ultimate threshold result. Method iterations only require multiplication and addition operations on the stored values making the method well suited for implementation in fixed-point digital signal processors.
Claims20
A digital signal processor, said digital signal processor comprising:
- a computer readable medium;
- and
- program code within the computer readable medium that causes a data processing system to perform a method including:
- calculating a cumulative histogram of a signal;
- calculating a weighted histogram of the signal;
- calculating an objective function value without using divides;
- calculating the initial estimate threshold value;
- calculating the threshold;
- analyzing the signal with the threshold.
The digital signal processor of claim 2, where calculating an objective function without using divides comprises:
- calculating an objective function value (C1) comprising:
- A1=N−N(k)
- A2=M·N(k)
- B1=(k·A1+M(k))
- B2=N·M(k)+A2
- C1=2·N(K)·B1−B2
- where:
- N(k)=kth cumulative histogram
- M(k)=kth weighted cumulative histogram
- C1 is the result of the objective function
- k is an initial estimate threshold value
The digital signal processor of claim 13 where the initial estimate threshold value is calculated as Kmax.
A method for determining a threshold of a signal, said method comprising:
- calculating a cumulative histogram of the signal;
- calculating a weighted histogram of the signal;
- calculating an objective function value (C1) comprising:
- A1=N−N(k)
- A2=M·N(k)
- B1=(k·A1+M(k))
- B2=N·M(k)+A2
- C1=2·N(K)·B1−B2
- where:
- N(n)=nth cumulative histogram
- M(n)=nth weighted cumulative histogram
- C1 is the result of the objective function
- k is an initial estimate threshold value
- calculating the initial estimate threshold value;
- calculating the threshold;
- analyzing the signal with the threshold.
Dependencies are inferred from the claim text.