EP3488524A1 - Digital filter with confidence input - Google Patents
Digital filter with confidence inputInfo
- Publication number
- EP3488524A1 EP3488524A1 EP17745153.1A EP17745153A EP3488524A1 EP 3488524 A1 EP3488524 A1 EP 3488524A1 EP 17745153 A EP17745153 A EP 17745153A EP 3488524 A1 EP3488524 A1 EP 3488524A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- input
- filter
- coefficients
- confidence
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H21/00—Adaptive networks
- H03H21/0012—Digital adaptive filters
- H03H21/0043—Adaptive algorithms
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H21/00—Adaptive networks
- H03H21/0012—Digital adaptive filters
- H03H21/0067—Means or methods for compensation of undesirable effects
Definitions
- the present disclosure relates to digital filters, in particular digital filters for noise suppression.
- sampling a signal (significantly) faster than its actual information content changes is a common practice that allows the enhancement of the digitized signal, exploiting the information's redundancy.
- Examples for such devices include capacitive-touch sensing or touchless position and gesture sensing systems, digital voltmeters, thermometers or pressure sensors.
- Exemplary capacitive sensing systems which can be subject to significant noise include the systems described in application note AN1478, “mTouchTM Sensing Solution Acquisition Methods Capacitive Voltage Divider”, and AN 1250, “Microchip CTMU for Capacitive Touch Applications”, both available from Microchip Technology Inc., the Assignee of the present application, and hereby incorporated by reference in their entirety.
- Another exemplary application is a touchless capacitive 3D gesture system - also known as the GestIC® Technology - manufactured by the Assignee of the present application.
- the sensor signals are typically subject to disturbance by various noise types, such as broadband noise, harmonic noise, and peak-noise.
- noise types such as broadband noise, harmonic noise, and peak-noise.
- the latter two can arise, for example, from switching power supplies, and are also addressed in electro -magnetic immunity standard tests, e.g. IEC 61000-4-4.
- the signal acquisition can also be interrupted in a scheduled or deterministic scheme; e.g., when multiplexing several sensors in time, or by irregular events such as data transmission failures. Such discontinuities or missing samples can cause undesired phase jumps in the signal. With digital filters designed for regular sampling intervals, this will corrupt the filter timing and can severely affect their noise-suppression performance. In analogy to erased messages in the context of channel coding in digital communications (Blahut, 1983; Bossert, 1999) we refer to missing samples and samples that do not carry useful information— e.g., due to peak noise— as Erasures.
- Fig. la shows a system 100 performing a basic procedure for estimating a noisy, real- valued baseband signal.
- the Analog-to-Digital Converter (ADC) 110 samples the signal at a rate (significantly) higher than its information changes.
- the digital signal is then input to a low- pass filter 120 and decimated with rate R by decimator 130.
- the downsampled result is processed further or is simply displayed, e.g. on a numeric display 140 as shown in Fig. la.
- the low-pass filter 120 is able to attenuate the higher frequency components of broadband noise, but will not thoroughly suppress noise peaks.
- An approach to suppress peak-noise, but still smoothing the input signal is a filter that averages over a subset of samples in a time window, excluding samples that have been identified as noise peaks or outliers, or excluding, for example, the n largest and the n smallest samples (Selective Arithmetic Mean (SAM) Filter or 'Sigma Filter' (Lee, 1983)).
- SAM Selective Arithmetic Mean
- FIR Finite Impulse Response
- a digital filter may comprise an assigned filter function with assigned filter coefficients, an input receiving input samples, another input receiving confidence values, and an output, wherein each input sample value is associated to an input confidence value and wherein each input sample is weighted with its associated confidence value, wherein the filter output depends on both the input samples and the input confidence values, and wherein the filter comprises accumulators configured to accumulate a predefined number of the confidence weighted input samples, the associated confidence values, the confidence values weighted with assigned filter coefficients, and the confidence weighted input samples further weighted with the assigned filter coefficients.
- the filter may comprising a first branch having a first accumulator receiving the input confidence values weighted with coefficients from a coefficient set and generating a first accumulated value; a second branch having a second accumulator receiving the input confidence values and generating a second accumulated value; a third branch having a third accumulator receiving input sample values weighted with coefficients from the coefficient set and the input confidence values and generating a third accumulated value; and a fourth branch having a fourth accumulator receiving the confidence weighted input values and generating a fourth accumulated value.
- the first accumulated value is subtracted from a constant value, wherein a result of the subtraction is being divided by the second accumulated value and multiplied with the fourth accumulated value, and added to the third accumulated value, and wherein the first, second, third, and fourth accumulator are subsequently cleared.
- a digital filter may comprise an assigned filter function with first and second filter coefficient sets, an input receiving input samples, another input receiving confidence values, and an output, wherein each input sample value is associated to an input confidence value; the filter output depends on both the input samples and the input confidence values, and wherein the digital filter comprises a first branch having a first accumulator receiving the input confidence values weighted with coefficients from the first coefficient set and generating a first accumulated value; a second branch having a second accumulator receiving the input confidence values weighted with coefficients from the second coefficient set and generating a second accumulated value; a third branch having a third accumulator receiving input sample values weighted with coefficients from the first coefficient set and the input confidence values and generating a third accumulated value; and a fourth branch having a fourth accumulator receiving the input values weighted with coefficients from the second coefficient set and the input confidence values and generating a fourth accumulated value.
- the first accumulated value is subtracted from a constant value, wherein a result of the subtraction is being divided by the second accumulated value and multiplied with the fourth accumulated value, and added to the third accumulated value, and wherein the first, second, third, and fourth accumulator are subsequently cleared.
- multiple instances of the filter are operated in parallel, each instance on a subset of input samples and associated confidence values, and with dedicated coefficients.
- a confidence value is represented by a digital logic value.
- the constant value is a sum of all coefficients.
- the assigned filter function is a low-pass filter function.
- the low-pass has been obtained from transforming a high-pass or band-pass to an equivalent low-pass domain.
- the assigned filter function has only positive valued coefficients or only negative valued coefficients.
- the assigned filter function has at least one non-zero valued coefficient with a different magnitude than another non-zero coefficients.
- a digital filter's DC gain is constant or approximately constant.
- a filter system may comprise a first and second digital filter, each comprising an assigned filter function with assigned filter coefficients, an input receiving input samples, another input receiving confidence values, and an output, wherein each input sample value is associated to an input confidence value and wherein each input sample is weighted with its associated confidence value; wherein the filter output depends on both the input samples and the input confidence values, and wherein the filter comprises accumulators configured to accumulate the confidence weighted input samples, the associated confidence values, the confidence values weighted with assigned filter coefficients, and the confidence weighted input samples further weighted with the assigned filter coefficients; and a demultiplexer receiving an input signal and generating input samples for said first and second digital filter.
- the system may further comprise a first outlier detector receiving said input samples for said first digital filter and generating associated confidence values; and a second outlier detector receiving said input samples for said second digital filter and generating associated confidence values.
- input samples for the first digital filter are high samples and input samples for the second digital filter are low samples.
- a filter system may comprise a first and second digital filter, each comprising an assigned filter function with first and second filter coefficient sets, an input receiving input samples, another input receiving confidence values, and an output, wherein each input sample value is associated to an input confidence value; the filter output depends on both the input samples and the input confidence values, and wherein each of the digital filter further comprises a first branch having a first accumulator receiving the input confidence values weighted with coefficients from the first coefficient set and generating a first accumulated value; a second branch having a second accumulator receiving the input confidence values weighted with coefficients from the second coefficient set and generating a second accumulated value; a third branch having a third accumulator receiving input sample values weighted with coefficients from the first coefficient set and the input confidence values and generating a third accumulated value; and a fourth branch having a fourth accumulator receiving the input values weighted with coefficients from the second coefficient set and the input confidence values and generating a fourth accumulated value; and wherein the system further comprises
- the system may further comprise a first outlier detector receiving said input samples for said first digital filter and generating associated confidence values; and a second outlier detector receiving said input samples for said second digital filter and generating associated confidence values.
- input samples for the first digital filter are high samples and input samples for the second digital filter are low samples.
- a digital filter may comprise an assigned filter function with assigned filter coefficients, an input receiving input samples, another input receiving confidence values, and an output, wherein each input sample value is associated to an input confidence value; wherein the filter output depends on the input samples, the input confidence values as well as the filter coefficients; wherein the filter contains multiple accumulators; wherein an output sample is produced after a predetermined number of sample values wherein associated confidence values have been input to the filter.
- a method for filtering digital input samples may comprise the steps of: receiving digital input sample values and associated input confidence values; accumulating the input confidence values weighted with coefficients from a coefficient set and generating a first accumulated value; accumulating the input confidence values and generating a second accumulated value; accumulating the input sample values weighted with coefficients from the coefficient set and the input confidence values and generating a third accumulated value; accumulating the confidence weighted input values and generating a fourth accumulated value.
- the method may further comprise subtracting the first accumulated value from a constant value, wherein a result of the subtraction is being divided by the second accumulated value and multiplied with the fourth accumulated value, and added to the third accumulated value, and subsequently clearing the first, second, third, and fourth accumulator.
- the constant value is a sum of all coefficients.
- the input confidence value is binary.
- a method for filtering digital input samples may comprise the steps of: receiving digital input sample values and associated input confidence values; accumulating the input confidence values weighted with coefficients from a first coefficient set and generating a first accumulated value; accumulating the input confidence values weighted with coefficients from a second coefficient set and generating a second accumulated value; accumulating input sample values weighted with coefficients from the first coefficient set and the input confidence values and generating a third accumulated value; and accumulating the input values weighted with coefficients from the second coefficient set and the input confidence values and generating a fourth accumulated value.
- the method may further comprise subtracting the first accumulated value from a constant value, wherein a result of the subtraction is being divided by the second accumulated value and multiplied with the fourth accumulated value, and added to the third accumulated value, and subsequently clearing the first, second, third, and fourth accumulator.
- the constant value is a sum of all coefficients.
- the input confidence value is binary.
- Fig. la shows exemplary acquisition of an analog signal, with analog-to-digital conversion and conventional noise suppression
- Fig. lb shows magnitude responses of different low-pass filters
- Fig. 2 shows typical tap weights of a low pass filter with finite impulse response
- Fig. 3 shows an exemplary block diagram of a data source and associated confidence generation as an input source for a digital filter
- Fig. 4 shows an exemplary implementation of a digital filter with confidence input
- Fig. 5 shows a system with an external confidence generating controller
- Fig. 6 shows an exemplary shift register implementation of a digital filter with confidence input
- Fig. 7 shows an example of redistribution of erased coefficient weight according to various embodiments
- Fig. 7A shows another example of redistribution of erased coefficient weight according to various embodiments for two level signals
- Fig. 8 shows an example of peak-noise suppression performance
- Fig. 9 shows a comparison of filters' coefficients and magnitude spectra with and without erasures
- Fig. 10 shows an example of a redistribution of erased coefficient weight in a high pass filter embodiment
- Fig. 11 shows an embodiment of a non-touching gesture detection system using an alternating quasi-static electric field
- Fig. 12 shows an example of a two level measurement
- Figs. 13 shows an implementation for packet data processing with minimum buffer requirements
- Fig. 15 shows an implementation similar to Fig. 13, but with a differential stage after decimation;
- Fig. 16 shows an implementation similar to Fig. 13, but with a general redistribution function
- Fig. 17 shows an implementation with a general redistribution function, and a binary confidence input realized with switches
- Fig. 18 shows the splitting of original filter impulse response into two
- Fig. 19 shows an implementation using two filters with confidence input for amplitude modulated ADC output samples 3 ⁇ 4.
- a reliable estimate of a real-valued baseband signal e.g. a demodulated and downsampled GestIC ® signal
- a reliable estimate of a real-valued baseband signal can be obtained where the input signal is oversampled and noisy.
- GestIC ® devices such as MGC3030 or MGC3130 or newer designs are available from the Assignee of the present application.
- Figure 11 shows a typical embodiment, wherein controller 740 represents the GestIC ® device.
- a general description and design guide such as the "GestIC ® Design Guide" published online Jan., 15, 2015 is available from Microchip Technology Inc. and hereby incorporated by reference.
- the 3D-gesture detection system 700 shown in Fig. 11 provides for a transmission electrode 720 which may be formed by a frame structure as shown in Fig. 11 and a plurality of receiving electrodes 710a..d. However, the entire rectangular area under the receiving electrodes 710a..d may be used as a transmission electrode or such an electrode can also be split into a plurality of transmission electrodes.
- the transmission electrode(s) 720 generate an alternating electrical field.
- a gesture controller 740 receives signals from the receiving electrodes 710a.. d which may represent the capacitances between the receive electrodes 710a.. d and system ground and/or the transmission electrode 720.
- the gesture controller 740 may evaluate the signals and provide a processing system 730 with human device input information. This information may be 3D moving coordinates similar to the 2-D moving information generated by a computer mouse and/or include commands that are generated from detected gestures.
- a peak-noise detection system or a deterministic noise indicator.
- broad-band noise namely applying a frequency (low pass) filter.
- peak noise namely applying a median filter over a window of signal samples.
- each input sample is associated with a confidence value.
- This confidence value is indicating whether the associated sample is an Erasure or not - i.e., whether or not it is an actually missing sample or a sample that is known to not carry useful information.
- the confidence values are assumed to be known by some other means. Such means can include, for example, deterministic input, or Outlier Detection methods such as the Grubbs' Test (Grubbs, 1950), the Generalized Extreme Studentized Deviate (GESD) test, or the Hampel identifier (Hampel, 1974).
- confidence values are used— for example— as weights in a Least Squares regression for improved alpha matting (J. Horentrup, 2014).
- Fig. 2 shows the filter coefficients of a typical low-pass filter - or 'windowing' - function, here exemplary a normalized Hamming window of length 8.
- Each filter coefficient defines the weight of its associated tap in a tapped delay line implementation, also shown in Fig. 2, where each tap is aligned underneath its associated coefficient in the bar plot. Therefore, we use the terms 'filter coefficient' and 'tap weight' synonymously.
- the tapped delay line contains 7 consecutive delay stages z and 8 tap weights.
- Exemplary input samples are also shown in Fig. 2. Other sample structures may apply with less or more stages.
- Fig. 2 shows a typical weight/coefficient distribution of a low pass filter which in this example is formed by, e.g., seven consecutive delay stages z "1 .
- Other sample structures may apply with less or more stages.
- FIG. 3 shows a block diagram of an exemplary Digital Filter with Confidence Input 300 and its input signal sources.
- the Data Source 320 is generating the signal x with samples Xk at discrete time k.
- the confidence information can also origin from some external means which we refer to as External Indicator 310. This external indication can be seen as a deterministic confidence input.
- Figure 5 shows a system 500 with a 2D capacitive touch detection and finger tracking system, as it is for example used in touch panels or touch displays, consisting of number of sensor electrodes "2D Electrode Pattern” 520 and a controller unit "2D Touch Controller” 510.
- Around the 2D Electrode Pattern are arranged four further electrodes A, B, C, D used with a 3D Gesture Controller 530 to form a capacitive 3D gesture detection system.
- the 2D touch detection system 510, 520 is active, it is interfering the received signals of the 3D gesture detection system, i.e. the received data of the 3D gesture detection system is noisy and unusable.
- the filter output signal y is
- the erased filter weight must be distributed onto the other filter coefficients.
- the preferred approach to do so is to distribute the erased weight
- the erased weight as shown in the right of the second plot, is evenly re-distributed onto the coefficients assigned to non-erased input samples, yielding Wi(k).
- the added on weights are shown differently hatched in the third plot.
- the coefficient weights are shown in the bottom shift register filter diagram as wo-w 7 in this embodiment.
- FIG. 8 An example for the filter's noise suppression performance is shown in Figure 8.
- the top plot shows the filter input signal, which is a slowly varying information signal with additive Gaussian noise, several noise peaks and - starting at sample index 250 - with additive 60Hz sinusoidal noise.
- the second plot shows that traditional low-pass filtering with a Hamming function of length 64 reduces the higher frequency noise, but only smears the noise peaks that are present in the input signal. However, having identified the noise peaks, they are suppressed completely according to various embodiments.
- Fig. 8 demonstrates the benefit of using a Hamming Erasure filter function instead of Moving Average: Compared to simple averaging over non-peak samples, i.e. Selective Arithmetic Mean filtering, the Hamming Erasure filtering yields better suppression of the broad-band noise contained in the input signal, yielding a smoother output.
- Fig. 9 illustrates how Erasures affect the filter's frequency response.
- the left side shows a typical low pass filter using a rectangular window and a Hamming window and its associated magnitude spectrum.
- the right side shows the same filtering using two erased samples. It can be observed that the spectrum of the Hamming Erasure filter is similar to the Rectangular erasure filter - depending on positions of Erasures.
- Equation (1.2) General Redistribution Function For binary confidence input, in Equation (1.2) the erased weight is evenly distributed onto the other coefficients.
- the two terms in (1.2) can be interpreted as two parallel filter branches whose output is summed-up.
- the filter function in the first term is computed from b and the confidence input, and the second term is a time -variant averaging filter.
- the latter can be generalized by introducing another FIR filter function g with coefficients gi, yielding
- ⁇ ' refers to a standard FIR filter with filter function b, and analog for the block denoted G' i
- the block denoted - refers to the division of 1 by the block's input data, i.e. the output of this block is the multiplicative inverse (reciprocal) of the input.
- the filter coefficients of the four filter blocks ( ⁇ ' and "G') are constant.
- the filters ⁇ ' and " G' processing the same input data, i.e.
- the adaptivity is contained in the filter blocks' input signals.
- the implementation is equivalent to a single FIR filter with adaptive filter coefficients w t (/c) .
- a characteristic property of the tapped delay line implementation of the FIR filter in Fig. 6 is that the tapped delay lines for the confidence values TDL-C and the tapped delay line for confidence-weighted input data TDL-XC are identical, i.e. they have the same number of delay stages, and the same tap weights b Q , b , ... and g 0> gi > — are connected to the respective delay stages.
- one or the other delay line may be simplified depending on the input variable type, e.g. binary confidence input.
- the delay lines or associated computation blocks can be simplified. Further, it does not matter if the weights b 0 , b x , ...
- TDL-C differs from the weights b 0 , b x , ... in TDL-XC by a constant factor, neither if the weights g 0 , g lt ... in TDL-C differ from the weights g 0 , g lt ... in TDL-XC by a constant factor, because such factors can be compensated for outside the tapped delay lines.
- the respective tap weights can also be set to 1, hence saving multiplications, and only the sum at the end of the tapped delay line before the (1/x) block is divided by 8, which can also be done by means of bit shift operations.
- conditional statements can also be allocated at each tap of the delay line: A tap- weighted input value b t ⁇ x k _ t or g i ⁇ x k _ t is then only added to the respective delay line's output sum value if the associated c fe _ £ is One.
- the samples x k can be directly input to TDL-XC and to not need to be multiplied with c k beforehand. The analog holds for TDL- C.
- the order of the filters with impulse responses b and g do not necessarily have to be equal. Without loss of generality the filters are defined to have equal order N, where N is at least as large as the maximum of the orders of the filters with b and g, and unused coefficients are assumed to be zero-valued.
- FIG. 12 shows exemplary measurement values with such high and low levels.
- An example is amplitude modulation (AM) with synchronous sampling of the analog received signal at twice the carrier frequency, where the information is contained in the difference between the 'high' and 'low' signal level.
- AM amplitude modulation
- This method is applied, for example, in capacitive touch detection systems, or GestIC ® Technology.
- the measurement (or 'received') signal of such an AM sensor system can, for example, be demodulated by alternatingly multiplying it with +1 and -1 , and then low- pass filtered in order to estimate the DC (zero-frequency) value— the actual information, the 'averaged' difference between the 'high' and 'low' samples.
- an input signal's samples must be sorted into sets of samples with the same expected value, and the digital filtering with confidence input—i.e. the re-distribution of coefficient weight— must happen in such a way that the all-over weight of the coefficients assigned to each set remains constant, which is most easily accomplished when re-distributing weight erased in one set within the same set.
- Fig. 7A shows an example windowing and DC computation for signals with two levels. Handling of erased values is performed individually for samples at 'high' and 'low' signal level, respectively as shown in Fig. 7A. Every other filter coefficient is assigned to a measurement value from the 'high' or the "low' signal level.
- Graph a) in Fig. 7A shows the original filter coefficients.
- Graph b) separates the 'high' level coefficients and graph c) the 'low' level coefficients.
- Graph d) in Fig. 7A shows that coefficient 3 corresponds to a sample at 'low' signal level and is erased. Its weight is re-distributed onto the other coefficients assigned to samples with a 'low' signal level.
- Graph e shows the redistribution with respect to the 'low' signal level coefficients.
- the bottom graph f) shows a combination of the redistributed 'low' signal level coefficients and the 'high' signal level coefficients. The expected output value is thus preserved.
- This method can be implemented with two data branches, one for samples with 'high' signal level and one for samples with 'low' signal level, summing the branches' outputs in the end. Again, as this is a time-variant filter, the re- distribution of filter coefficients is updated for each output sample.
- a measure that fulfills (1.4) and is readily available is the sum of the filter coefficients being weighted with their corresponding input confidence values, i.e.
- this output can be used for high-level control, e.g. 'do not trigger touch event if output confidence is too low'.
- the proposed approach is applicable to any low-pass FIR filter.
- all filter coefficients should have the same sign, e.g. be positive valued.
- the requirement of a constant DC filter gain can also be fulfilled when (some of the) tap weights are negative valued.
- the coefficients of a Rectangular window, Triangular window, Hamming and Hann window are in compliance with these rules.
- Fig. 10 shows an example of a high- pass filter whose original filter coefficient weights alternate between positive and negative sign as shown in the first plot from the top.
- the high pass filter coefficients are demodulated using and alternating sign as shown in the second plot from the top.
- the same method as with the low pass filter shown in Fig, 2 is applied as shown in the third and fourth plot from the top.
- the modified weights as shown in the fourth plot from the top are then re -modulated using the inverse alternating signs. This results in distributed weights as shown in the fifth plot from the top.
- the filter's input signal can be demodulated, too, and filtered with the equivalent low-pass with coefficients according to the forth plot from the top, before modulating the signal again. No matter if the input signal is filtered with the high-pass filter directly, or demodulated and being filtered with the equivalent low-pass filter, it is important that the samples of the high pass filter's input signal - if it would be demodulated - would have a single expected value.
- the proposed concept is applicable to any filtering system where the input signal is sampled faster than the actual information changes - the higher the sampling rate, the better.
- such systems include 3D capacitive sensor systems, such as Assignees GestIC system and 1D/2D capacitive touch solutions.
- the filtering method may further more be applied to other sensor signals and is not restricted to capacitive sensor systems.
- the filter When initializing the filter with arbitrary, but zero-confidence, data, it provides an estimate of the input signal from turn-on time. Hence, the filter does not show the typical step response when filtering a signal with a non-zero mean and the filter conditions have been initialized with zeroes.
- Fig. 13 to Fig. 17 show yet other implementations of an Erasure Filter for packet data processing. These embodiments use a different, yet very memory efficient solution.
- each input sample is contributing only to a single output value, and hence the implementation can be done using accumulators acc01..acc04 instead of tapped delay lines (buffers) - as shown in the embodiments according to Figs. 13-17.
- these accumulators are set to Zero, and then the respective input values are successively added to them.
- L the length of a data packet
- N the order of a low-pass filter
- Cke[0,l] i.e. 0 ⁇ Ck ⁇ l
- Ck 0 means that Xk is considered— e.g. by an peak noise or outlier detector— to not carry any useful information and shall not contribute to the filter output
- Figure 13 shows an implementation for Packet Data Processing with Minimum Buffer
- Fig. 13 also includes the multiplicative inversion denoted by — i.e., the block's output value is 1 divided by the input value.
- bN-k in Figure 13 can also be replaced by bk.
- the multiplications with Ck i.e. the weighting with the confidence values— can also be implemented with switches, cf. Figure 14 or a conditional increment of a variable
- the filter is no longer an FIR filter but can be an IIR filter.
- the sum over the L input values can also be done by remembering the values of accOl, acc02, acc03 and acc04 before inputting the data for the current packet, and to subtract these remembered values from the respective values of accOl, acc02, acc03 and acc04 after inputting the data for the current packet.
- This can, for example, be realized by employing a first order CIC filter as for example known from "An Economical Class of Digital Filters for Decimation and Interpolation", by Eugene B.
- sample Values with Multiple Expected Values In many applications, e.g. capacitive sensing, the actual information is amplitude modulated, and the ADC output data needs to be demodulated prior to low-pass filtering, and both the ADC output values and the demodulated ADC output values have two expected values.
- the Digital Filter with Confidence Input can only be directly applied to input samples with a single expected value.
- the demodulated ADC output values need to be split into two sets with a single expected value for all samples within each set.
- the ADC alternatingly outputs samples at two different expected signal levels, denoted as Low and High samples, respectively, and we assign even time indices k to Low samples, and odd time indices to High samples.
- the original low-pass filter impulse response is split into two by separating coefficients bi with even i and odd i.
- Figure 19 shows how the demodulated samples Xk are distinguished for even and odd values of k and split onto two data branches, and with q being the floor value of k/2, two instances of the Digital Filter with Confidence Input according to Fig. 13 - but with differing filter impulse responses, cf.
- Figures 18 and 19— are employed to process samples Xk with even and odd indices k, respectively.
- Each data branch in this example has its own peak noise detector generating the confidence values c.
- the above mentioned digital filters can be formed by hardware, for example a programmable logic device or software, for example in a microcontroller, processor or digital signal processor.
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Complex Calculations (AREA)
- Analogue/Digital Conversion (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/215,930 US10277203B2 (en) | 2015-02-27 | 2016-07-21 | Digital filter with confidence input |
PCT/EP2017/068172 WO2018015412A1 (en) | 2016-07-21 | 2017-07-18 | Digital filter with confidence input |
Publications (1)
Publication Number | Publication Date |
---|---|
EP3488524A1 true EP3488524A1 (en) | 2019-05-29 |
Family
ID=59409322
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP17745153.1A Pending EP3488524A1 (en) | 2016-07-21 | 2017-07-18 | Digital filter with confidence input |
Country Status (6)
Country | Link |
---|---|
EP (1) | EP3488524A1 (en) |
JP (1) | JP6913099B2 (en) |
KR (1) | KR102431556B1 (en) |
CN (1) | CN108702143B (en) |
TW (1) | TWI721194B (en) |
WO (1) | WO2018015412A1 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2022132212A1 (en) * | 2020-12-16 | 2022-06-23 | Microchip Technology Incorporated | Methods and systems for determining a noise-robust acquisition configuration for operating a sensor system |
US11916582B2 (en) | 2020-12-16 | 2024-02-27 | Microchip Technology Incorporated | Methods and systems for determining a noise-robust acquisition configuration for operating a sensor system |
CN114485723B (en) * | 2021-02-08 | 2024-02-27 | 北京理工大学 | High-rotation body air alignment method of self-adaptive robust matrix Kalman filtering |
Family Cites Families (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4393456A (en) * | 1981-03-19 | 1983-07-12 | Bell Telephone Laboratories, Incorporated | Digital filter bank |
US5231680A (en) * | 1990-12-21 | 1993-07-27 | Ampex Systems Corporation | Horizontal adaptive error concealment system for component digital television signals |
TW247984B (en) * | 1990-12-21 | 1995-05-21 | Ampex | |
US5249578A (en) * | 1992-09-15 | 1993-10-05 | Hewlett-Packard Company | Ultrasound imaging system using finite impulse response digital clutter filter with forward and reverse coefficients |
JPH08137608A (en) * | 1994-11-07 | 1996-05-31 | Matsushita Electric Ind Co Ltd | Three-dimension coordinate input device |
JP3181879B2 (en) * | 1998-07-22 | 2001-07-03 | シャープ株式会社 | Transmission device |
US8671128B1 (en) * | 2006-06-21 | 2014-03-11 | Ikanos Communications, Inc. | Method and apparatus for a finite impulse response filter |
US7865030B2 (en) * | 2006-09-13 | 2011-01-04 | Broadcom Corporation | Method and system for motion compensated temporal filtering using both FIR and IIR filtering |
TW200906170A (en) * | 2007-07-18 | 2009-02-01 | Sunplus Technology Co Ltd | Image noise estimation system and method |
US8660220B2 (en) * | 2008-09-05 | 2014-02-25 | Lsi Corporation | Reduced frequency data processing using a matched filter set front end |
US8885890B2 (en) * | 2010-05-07 | 2014-11-11 | Microsoft Corporation | Depth map confidence filtering |
EP2463856B1 (en) * | 2010-12-09 | 2014-06-11 | Oticon A/s | Method to reduce artifacts in algorithms with fast-varying gain |
GB2487361A (en) * | 2011-01-17 | 2012-07-25 | Sony Corp | A digital interpolating FIR filter using fewer multipliers |
CN104463819B (en) * | 2013-09-20 | 2019-03-08 | 汤姆逊许可公司 | Image filtering method and device |
EP2887309A1 (en) * | 2013-12-20 | 2015-06-24 | Thomson Licensing | Method and apparatus for filtering an image |
-
2017
- 2017-07-18 JP JP2018540147A patent/JP6913099B2/en active Active
- 2017-07-18 EP EP17745153.1A patent/EP3488524A1/en active Pending
- 2017-07-18 KR KR1020187024996A patent/KR102431556B1/en active IP Right Grant
- 2017-07-18 CN CN201780013208.9A patent/CN108702143B/en active Active
- 2017-07-18 WO PCT/EP2017/068172 patent/WO2018015412A1/en active Application Filing
- 2017-07-20 TW TW106124396A patent/TWI721194B/en active
Also Published As
Publication number | Publication date |
---|---|
CN108702143B (en) | 2023-08-01 |
KR102431556B1 (en) | 2022-08-11 |
TW201806316A (en) | 2018-02-16 |
JP2019525502A (en) | 2019-09-05 |
KR20190031429A (en) | 2019-03-26 |
JP6913099B2 (en) | 2021-08-04 |
WO2018015412A1 (en) | 2018-01-25 |
CN108702143A (en) | 2018-10-23 |
TWI721194B (en) | 2021-03-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10135424B2 (en) | Digital filter with confidence input | |
Jansen | Noise reduction by wavelet thresholding | |
Saxena et al. | Removal of powerline interference from ECG signal using FIR, IIR, DWT and NLMS adaptive filter | |
Smolka et al. | Adaptive rank weighted switching filter for impulsive noise removal in color images | |
Dean | Frequency analysis for gravity and magnetic interpretation | |
Ferrer et al. | Fourier and Walsh digital filtering algorithms for distance protection | |
EP3488524A1 (en) | Digital filter with confidence input | |
US10277203B2 (en) | Digital filter with confidence input | |
Pecora et al. | Discontinuous and nondifferentiable functions and dimension increase induced by filtering chaotic data | |
Mercorelli | A denoising procedure using wavelet packets for instantaneous detection of pantograph oscillations | |
Yang et al. | Trend extraction based on separations of consecutive empirical mode decomposition components in Hilbert marginal spectrum | |
Meinl et al. | A nonlinear filtering algorithm based on wavelet transforms for high-frequency financial data analysis | |
Furlani Bastos et al. | Detection of inconspicuous power quality disturbances through step changes in rms voltage profile | |
Alhaj et al. | Power system harmonics estimation using sliding window based LMS | |
Ferreira et al. | Enhancement of the comb filtering selectivity using iterative moving average for periodic waveform and harmonic elimination | |
Zhang et al. | Recursive parametric frequency/spectrum estimation for nonstationary signals with impulsive components using variable forgetting factor | |
Isabelle et al. | Effects of convolution on chaotic signals. | |
Bhoi et al. | Multidimensional approaches for noise cancellation of ECG signal | |
Zaw et al. | Performance Comparison of Noise Detection and Elimination Methods for Audio Signals | |
CN112039494B (en) | Low-pass filtering method, device, equipment and medium for overcoming azimuth zero crossing | |
Heim et al. | Digital filter with confidence input | |
Liang et al. | Adaptive frequency estimation with low sampling rates based on robust Chinese remainder theorem and IIR notch filter | |
Kuang et al. | MSSA-based adaptive low-frequency noise reduction using spectrum overlap measure | |
Radhakrishnan et al. | Fault tolerant Fermat Number Transform domain adaptive filters based on modulus replication RNS architectures | |
Wells et al. | Short-Time Wavelet Analysis of Analytic Residuals for Real-Time Spectral Modelling |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20190219 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
AX | Request for extension of the european patent |
Extension state: BA ME |
|
DAV | Request for validation of the european patent (deleted) | ||
DAX | Request for extension of the european patent (deleted) | ||
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
17Q | First examination report despatched |
Effective date: 20210531 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |