JPH04306048A - Jitter evaluation method in picture input - Google Patents

Abstract

PURPOSE:To evaluate a deteriorated picture read by a reader directly and to grasp a 2-dimension distribution of a jitter component in the picture. CONSTITUTION:A reference picture is read as a digital picture data. A sinusoidal reference grating picture XY of a single frequency is used as an input reference picture as a method to evaluate jitter from the picture input device. Fourier transformation is obtained in the direction of the spatial frequency distribution of the reference grating with respect to the inputted picture XY, the picture is segmented by a filter function having a band width equivalent to the spatial frequency of the reference grating, shifted in a direction of DC, subject to Fourier inverse transformation and a phase modulation component is extracted to calculate jitter of the picture input device.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for evaluating jitter in an image input device, and more particularly to a measurement method and signal processing method for measuring jitter in an image input device, including its spatial distribution. For example, it is applied to an evaluation method of an image input device and an image signal measurement.

0002

2. Description of the Related Art Generally, in image input devices such as digital copying machines and image scanners, for example, as shown in FIG.
In a type called a reduction type, two traveling bodies with mirrors called a first traveling body 11 and a second traveling body 12 scan an image original 13, and the image is formed on a light receiving element 15 by an imaging lens 14. Moreover, as shown in FIG.
In the 1-magnification imaging type, one traveling body 21 having a 1-magnification imaging element 22 and a 1-magnification light receiving element 23 scans an image original 24 . Usually these running bodies are motor driven by wires. When these wheels run, mechanical interaction between wires, motors, bearings, etc. causes vibrations in running. This is jitter. This jitter has a direct effect on the read image.

For example, as shown in FIGS. 11(a) and 11(b), a phenomenon in which the edges are wavy when reading diagonal edges is a typical example of noticeable jitter. In order to analyze these phenomena, we investigated the running speed fluctuations caused by magnetic tape, laser Doppler vibrometers, laser interferometers,
Measurements were carried out using optical and magnetic linear encoders. Regarding this point, for example, JP-A-1-1
It is described in JP-A No. 62114 and JP-A-1-161114. The HP system is a typical laser interferometer.

The method using magnetic tape will be briefly explained with reference to FIG. This involves fixing one end 32 of the magnetic tape to a running body 31, passing the other end 33 through a magnetic head 34, and measuring the speed at which pulses are written at the end of the magnetic tape as the running body travels. This measurement method is a one-dimensional measurement of the position where the magnetic tape 32 is fixed. In other words, the speed at one point on the traveling object is merely measured. Furthermore, although the measurement accuracy is improved in a system using a laser interferometer, it is essentially a one-dimensional measurement similar to that of a magnetic tape, and the system itself is very expensive. Furthermore, both methods are indirect measurements of vibration, and do not directly measure and evaluate images degraded by jitter. In reality, the image signal affected by jitter is a two-dimensional signal, and two-dimensional analysis of the location where jitter occurs is desired.

[0005]

[Purpose] The present invention has been made in view of the above-mentioned circumstances, and aims to directly evaluate a deteriorated image read by a reading device and to grasp the two-dimensional distribution of jitter components within the image. The purpose of this invention is to provide a method for evaluating jitter in an image input device as described above.

[0006]

[Structure] In order to achieve the above object, the present invention provides an evaluation method for evaluating jitter of an image input device by reading a reference image as digital image data and performing arithmetic processing on the input reference image. , arithmetic processing is performed using a method using a single frequency sine wave reference grid image as the input reference image, Fourier transform is obtained for the input reference image in the spatial frequency distribution direction of the reference grid, and the reference grid is The line spectrum is cut out using a filter function that has a center frequency at a spatial frequency of It is characterized in that it is shifted in the DC direction by an amount equivalent to the frequency, and then subjected to inverse Fourier transform to extract a phase modulation component and calculate the jitter of the image input device. Hereinafter, the present invention will be explained based on examples.

First, FIG. 1 shows an input reference image used as a reference. As shown, the reference image has a constant spatial frequency distribution in a single direction. The recorded image signal is a sine wave signal. The image input device to be evaluated is, for example, an image scanner or a digital copying machine, as shown in FIG. It is meant to move to. As shown in FIG. 2, the reference grid image is placed so that the reference grid is perpendicular to the main scanning direction of the image input device, and the image is read. If there is a component in the main scanning direction (x direction) in the jitter of the moving body of the image input device, the input image will be distorted in the phase of the reference grating at the jitter occurrence locations A and B, as shown in FIG. The amplitude of jitter can be calculated by detecting the phase shift of .

Next, a method for calculating the phase shift of the reference grating will be described. For example, in the case of a 400 dpi A3 image input device, if the x direction in FIG. 3 is taken as the short side of the A3, and the y direction is taken as the long side of the A3, the number of pixels Nx Ny is approximately 4600×
It will be about 6,600. Here, the arrangement of pixels along the x direction is defined as a main scanning line. The reference grid image is arranged such that the spatial frequency distribution direction of the grid is along the x-axis. Here, the spatial frequency of the reference grid is d (line pare
/inch). The number M of pixels constituting one period of spatial frequency is M=400/d (1).

By the way, if the reference grid is a normalized sine wave with distribution in the x direction, the image density distribution Iref(x,y) of the reference grid image is Iref(x,y)=1+cos(2πdx)...
It is expressed as (2). Also, if the x-direction jitter component at points x, y is J(x, y), then the image density distribution I(x, y) read by the image input device to be evaluated is
I(x,y)=a(x,y)+b(x,y)cos
(2πdx+φ(x,y)) …(3) However, φ(x
, y)=2πdJ(x, y) where a(x, y), b(
x, y) are factors due to disturbances other than the x-direction jitter component of the image input device. For example, a(x, y) is illumination unevenness, b(x, y) is sensitivity variation of the light receiving element, etc. All of these are unnecessary components for evaluating jitter and are unknown functions. From the above equation (3), a(x
, y), b(x, y) and extracting only the information on φ(x, y).

Next, a procedure for extracting a desired signal component φ(x,y) from the read image density distribution I(x,y) expressed by equation (3) will be explained. Expression (3) is expressed in complex form as follows.

[0011]

[Equation 1] However, c (x, y) = (1/2) b (x, y) exp (
jφ(x,y))* is the complex conjugate, j is the imaginary unit, and the one-dimensional Fourier transform I(fx,y) for the variable x in equation (4).
think of. i.e.

0012

[Math 2]

[0013] Here, A and C are a(x, y), c(x, y
) in the x direction. The first term A is a spectrum caused by signal components due to uneven illumination and variations in sensitivity of the light-receiving element, and is distributed mainly around the DC component. The second and third terms are due to the line spectrum of the reference grid spatial frequency, and are jitter components J(x,
y) and has a finite spread around d. These three terms are completely separated in the spatial frequency spectral domain if the spatial frequency d of the reference grid is high enough.

FIG. 4 shows an example in which the reference grid read image is distorted due to jitter. In the figure, ■ is a main scanning line with distortion due to jitter, and ■ is a main scanning line without distortion. Figures 5(a) and (b) show reference grid read image signals in the main scanning direction at positions with distortion and positions without distortion, and Figure 5(a) is based on the main scanning line of ■ in Figure 4. An image signal, that is, a signal portion that has undergone phase modulation due to jitter is shown. Figure 5(b) shows ■ in Figure 4.
This is an image signal based on the main scanning line of , that is, an image signal without distortion.

FIGS. 6(a) and 6(b) are diagrams in which the spread of the spectrum shows spatial distribution information of the jitter component, and are diagrams showing the Fourier transform spectra of image signals with and without jitter. It is. FIG. 6(a) is a Fourier transform spectrum in the case of ■ in FIG. 4 with jitter, and FIG. 6(b) is a diagram showing a Fourier transform spectrum in the case of ■ in FIG. 4 without jitter. . In Fig. 6(a), there is a broadening of the line spectrum, whereas
In FIG. 6(b), there is no expansion.

Next, the Fourier transform spectrum of equation I(x,y)(4) is applied with a bandpass filter shown in FIG. 7(a). This is to cut out the Fourier transform spectrum of formula I(x,y) (4) over an appropriate width centered around frequency d, including the finite spread of the line spectrum due to phase modulation (Figure 7(b) ). FIG. 7(a) shows a simple rectangular filter function. In this operation, the rectangular function is multiplied in the spatial frequency domain, so in real space the image signal is s
It will undergo a convolution operation with the inc function. Generally, the sinc function has wide side lobes, so it is smoothed by convolution in real space. In the field of digital signal processing, many functions with narrow side lobes in real space have been proposed.
Hanning function, Hamming function, black
The man function etc. are known. In the above explanation, the rectangular function was described as a filter function to simplify the explanation, but in reality, when considering the spatial resolution when returning to real space, it is effective to use the narrow sidelobe function mentioned earlier. .

[0017] As a result, only the second term of equation (5) is extracted. At this stage, low frequency components such as illumination unevenness and sensitivity variations of the light receiving elements included in the first term A of equation (5) are removed. This spatial frequency filtering operation is very effective, for example, when the reference grid is not a perfect sine grid, or when noise due to dust or electrical spikes is present during image reading. That is, since the deterioration of the image signal due to these noises occurs in a high frequency region compared to the reference high frequency, it does not affect the spectrum shape near d. Furthermore, even if the reference grid is, for example, a rectangular grid,
The effect is only that the spectrum spreads in the high range, so
It will be removed by a similar filtering operation. That is, the present invention is an epoch-making method that additionally has the property of being resistant to this type of noise.

Next, the extracted Fourier transform spectrum is shifted d to the left along the frequency axis, and the line spectrum based on the reference grid is moved to the DC position (FIG. 7(c)).
. A shift in the frequency plane (frequency transition) corresponds to a phase rotation in real space. With this operation, the rotation of the phase due to the first phase term of the cos equation (3) is returned to the original state. Here, equation (3) is expressed as follows, that is, jitter component J(
It can be regarded as a signal in which a phase carrier of frequency d is superimposed on a function having phase terms (x, y). Removing the first phase term in equation (3) cos corresponds to removing the phase carrier of the phase modulated signal. By this operation, I(x
, y) C(fx, y) could be extracted from equation (4).

Here, C(fx,y) is the Fourier transform of c(x,y). Therefore, the Fourier inverse transform operator is F
-1, F-1[C(fx,y)]=c(x,y)
=(1/2)b(x
,y)exp(jφ(x,y))...(6) Furthermore, the operators that take the real and imaginary parts of complex numbers are Re[], Im[
], then for the right side of (6)

[0020]

[Math 3]

[0021] Therefore

[0022]

[Math 4]

From (3), only the jitter component is a(x
, y) can be extracted without being affected by b(x, y). The flow of the above processing is shown in FIG. Below, each step will be explained in order. Step 1: Perform image input. Step 2: Next, set n=1. step 3: One-dimensional Fourier transform for the nth line (
FFT). Step 4: Filter in the frequency domain. Step 5: Shift the line spectrum to the DC position. Step 6: Perform one-dimensional inverse Fourier transform. Step 7: Check whether n=N. Step 8: If n=N, set n=n+1 and return to step 2. Step 9: If n=N, it is stored as display data.

[0024]

[Effects] As is clear from the above description, the present invention has the following effects. (1) Effect corresponding to claim 1: It becomes possible to independently and quantitatively detect the spatial distribution of the traveling body jitter in the image of the image input device at each point. Moreover, even if the distortion of the reference grid image is less than the reference grid period, it can be sensitively detected. Furthermore, processing can be performed using only software, without requiring expensive hardware such as a laser interferometer.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an input reference image for explaining an embodiment of a jitter evaluation method in an image input device according to the present invention.

FIG. 2 is a diagram showing an image input device to be evaluated.

FIG. 3 is a diagram showing the occurrence of jitter.

FIG. 4 is a diagram showing a reference grid image distorted by jitter.

FIG. 5 is a diagram showing a reference grid read image signal.

FIG. 6 is a diagram showing a Fourier transform spectrum of an image signal.

FIG. 7 is a diagram showing how the extracted Fourier transform spectrum is moved to a DC position.

FIG. 8 is a flowchart for explaining a jitter evaluation method in an image input device according to the present invention.

FIG. 9 is a diagram showing a conventional image input device.

FIG. 10 is a diagram showing another conventional image input device.

FIG. 11 is a diagram showing an original image and an image affected by jitter.

FIG. 12 is a diagram showing an apparatus using a magnetic tape method for analyzing jitter phenomena.

Claims (1)

[Claims] Claim 1. An evaluation method for evaluating jitter of an image input device by reading a reference image as digital image data and performing arithmetic processing on the input reference image, wherein the input reference image is a single frequency image. Computation processing is performed using a method using a sine wave reference grid image,
A Fourier transform is obtained for the input reference image in the direction of the spatial frequency distribution of the reference grid, and the image is cut out using a filter function that has a center frequency at the spatial frequency of the reference grid and a bandwidth equivalent to the spatial frequency of the reference grid. , the line spectrum generated by cutting out, including the spread of the line spectrum,
A method for evaluating jitter in an image input device, characterized in that the jitter in the image input device is calculated by shifting the frequency domain in the direct current direction by an amount equivalent to the spatial frequency of a reference grating, and then performing inverse Fourier transform to extract a phase modulation component.