JP2833532B2 - Method and apparatus for removing noise from optical sensor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a noise elimination method and a noise elimination apparatus which are capable of automatically eliminating linear noise contained in image data obtained by an optical sensor.

[0002]

2. Description of the Related Art Various linear noise, vertical stripe noise, horizontal stripe noise, and check noise may appear in data acquired by an optical sensor. For example, as shown in FIG. 3, a CC used for reading an image by a facsimile or the like is used.
In the case of a D-line sensor, the linear noise described above is easily generated when a difference in sensitivity occurs between a plurality of sensors arranged in a line or when the CCD bias voltage fluctuates. In particular, among the linear noises described above, noise due to CCD bias voltage fluctuation becomes horizontal stripe noise that periodically extends in the main scanning direction of the CCD, that is, extends in the sub-scanning direction on the image.

[0003] In an optical sensor in which such linear noise occurs, attention is paid to a specific CCD element, and a frequency analysis of an output change accompanying image acquisition gives a power spectrum having a peak at a specific frequency as shown in FIG. Can be It has been found that this peak corresponds to linear noise, and therefore, it is possible to reduce noise by reducing or removing the component of this specific frequency in the frequency domain. However, these linear noises do not occur in all data and their intensities are not constant. Therefore, if such noise removal processing is performed on data with low noise, noise may be given instead.

[0004]

Therefore, there has been a situation in which noise processing cannot be executed uniquely in the prior art. Therefore, a frequency analysis of image data is performed to determine a frequency band containing noise and its intensity. Are individually checked, and noise removal processing is performed only when it is determined that noise removal is necessary. For this reason, there is a problem that human judgment is involved in noise removal, which makes it difficult to automate noise removal and that the processing time for noise removal becomes longer. SUMMARY OF THE INVENTION An object of the present invention is to provide a noise elimination method and apparatus capable of automatically determining noise elimination and executing suitable noise elimination.

[0005]

SUMMARY OF THE INVENTION A noise removing method according to the present invention is obtained from an optical sensor composed of a plurality of elements.
From the image data, the data of the same element is used as sample data.
And obtain a power spectrum of the obtained data.
Calculation and the required frequency of the power spectrum
Calculate power average value in several ranges for each required frequency shift
And determine whether the obtained average changes continuously
And detects a discontinuous portion where the average value does not change continuously.
The average value of the discontinuous part and the part before it
Is calculated, and the ratio is set to a preset threshold.
Execute noise processing when the value is larger than the value .

[0006] Further, the noise elimination device of the present invention comprises a plurality of noise elimination devices.
Data obtained from an optical sensor composed of various elements
From the same device as sample data
A sample data acquisition unit and the power of the acquired data
-A power spectrum calculator for calculating the spectrum;
Power average in the required frequency range of the power spectrum
A moving average calculation unit for calculating a required frequency shift amount,
If a discontinuous part is detected from the change state of the calculated average value,
Continuity judgment unit and discontinuous part when discontinuous part is detected
Ratio calculation unit that calculates the ratio of each average value of the previous part
And if this ratio is greater than a preset threshold
A determination unit that outputs a noise processing determination when the
Noise removal processing that performs noise processing based on logical judgment
Unit.

[0007]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the present invention. The optical sensor OS is configured as a CCD line sensor.
Is main-scanned in the line direction and sub-scanned in a direction perpendicular to the line direction to obtain two-dimensional data.
The sample data obtaining section 1 receives data from the CCD line sensor and extracts image data obtained by scanning the same element in the sub-scanning direction from the data.

[0008] The power spectrum calculator 2 performs FFT such as Fourier transform and the like on the extracted data to obtain a power spectrum. Moving average calculator 3
Is a moving average based on the power spectrum,
For example, the average value calculation range = 10 Hz and the moving amount = 5 Hz are calculated. At this time, if the data does not include a noise component, the power decreases smoothly as the frequency increases, so that the moving average becomes gradually smaller.

The discontinuity determination unit 4 sequentially compares changes in the moving average value from the low frequency side, and detects a frequency band in which the average value increases, that is, a discontinuous portion. Ratio calculation unit 5
Calculates the ratio between the discontinuous portion and the average value before the discontinuous portion, if any. The determination unit 6 calculates the calculated ratio,
The value is compared with a preset threshold value. The noise removal processing unit 7 performs a noise removal process based on the output of the determination unit 6. As described above, the noise removal processing is a processing method for reducing or removing a frequency component having a peak at a specific frequency in the power spectrum. Further, the distortion correction processing unit 8 corrects distortion occurring in the power spectrum of the image data.

A noise elimination method using the noise elimination device having the above configuration will be described with reference to the flowchart of FIG. Image data output from the CCD line sensor OS, which is an optical sensor, is input to the sample data acquisition unit 1, and image data when the same element is scanned in the sub-scanning direction is extracted from these data. The extracted image data is subjected to FFT and power conversion in the power spectrum calculation unit 2, and the power spectrum is obtained.
Then, based on the power spectrum, the moving average calculation unit 3 calculates a moving average based on the amount of movement in the above-described range.

Then, the discontinuity judging section 4 sequentially compares the change of the moving average value from the low frequency side, and detects a frequency band in which the average value increases, that is, a discontinuous portion.
At this time, if the data does not include a noise component, the power decreases smoothly as the frequency increases,
The moving average gradually decreases.

On the other hand, when there is a discontinuous portion, the ratio calculating section 5 calculates the ratio between the discontinuous portion and the average value before the discontinuous portion. Then, the determination unit 6 compares the calculated ratio with a preset threshold value. As a result of the comparison, if the ratio exceeds the threshold value, a processed signal is output to the noise removal processing unit 7, and the noise removal processing unit 7 reduces or removes the frequency component in the discontinuous portion to remove the noise. Execute the process.

When the discontinuity determining unit 4 determines that no discontinuous portion has occurred, or when the determining unit 6 determines that the ratio is equal to or less than the threshold value, it is determined that the noise amount is small, and the noise removal processing is performed. Is not executed, and only the distortion correction processing is executed in the distortion correction processing unit 8.

As described above, since the noise amount is detected and the noise removal is executed only when it is determined that the noise amount is large and the removal processing is necessary based on the noise amount, the processing time can be reduced. In addition, when the noise amount is small, the noise processing is not performed, so that the noise does not increase.

[0015]

The present invention described above, according to the present invention is, from the power spectrum of the sampled data acquired from the image data
The average power in the required frequency range is calculated as the required frequency shift
Calculated for each time, and the discontinuity
Part, and the ratio of the average value before and after the discontinuous part
Is compared with a predetermined threshold to detect noise, and a noise removal process is performed based on the noise detection result. When noise removal is required, the noise removal process is automatically performed. When noise removal is not required, it is possible to not perform the noise removal, and it is possible to quickly obtain image data with less noise.

[Brief description of the drawings]

FIG. 1 is a block diagram of a noise removing apparatus according to the present invention.

FIG. 2 is a flowchart of a noise removal method of the present invention.

FIG. 3 is a diagram for explaining linear noise in the optical sensor.

FIG. 4 is a power spectrum diagram including a noise component.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Sample data acquisition part 2 Power spectrum calculation part 3 Moving average calculation part 4 Discontinuity judgment part 5 Ratio calculation part 6 Judgment part 7 Noise removal processing part 8 Distortion correction processing part

Claims (4)

(57) [Claims] 1. An optical sensor comprising a plurality of elements.
Sample data of the same element from the image data obtained
Data, and the power of this
Calculate the spectrum and calculate the power spectrum
The average power in the required frequency range is calculated as the required frequency shift
Calculated for each case and whether the obtained average value changes continuously
And the discontinuous portion where the average value is not continuously changed
Is detected, each of the discontinuous part and the part before it is
Calculate the ratio of the average value and the ratio is set in advance
A noise removal method for an optical sensor , wherein noise processing is performed when the noise is greater than a threshold value . 2. The average value is calculated from a lower frequency to a higher frequency.
The average value obtained is gradually reduced.
The frequency domain in the state is determined as a continuous state, and the average
Request to detect the frequency domain where the value increases as a discontinuity
Item 2. The method for removing noise from an optical sensor according to Item 1. 3. A sample data acquisition section for acquiring data of the same element as sample data from image data obtained from an optical sensor composed of a plurality of elements, and a power spectrum for calculating a power spectrum of the acquired data. A calculating unit, a moving average calculating unit that calculates a power average value of a required frequency range of the power spectrum for each required frequency shift amount, and detecting a discontinuous portion from a change state of the calculated average value. A discontinuity determining unit, a ratio calculating unit that calculates a ratio of each average value of the discontinuous portion and a portion before the discontinuous portion when the discontinuous portion is detected, and a ratio calculating unit that calculates the ratio by a predetermined threshold value. A noise determination unit that outputs a noise processing determination when the noise is large, and a noise removal processing unit that performs noise processing based on the noise processing determination. Removal device. 4. The data from which noise has been removed when the noise processing determination is output, and the noise processing determination is output.
4. The apparatus according to claim 3, further comprising a distortion correction processing unit that performs a distortion correction process on each of the data that is not subjected to noise removal when there is no noise.
The noise removing device as described in the above .