WO1999067743A1

WO1999067743A1 - Image correcting method and image inputting device

Info

Publication number: WO1999067743A1
Application number: PCT/JP1999/003243
Authority: WO
Inventors: Yoshikazu Ichiyama
Original assignee: Yoshikazu Ichiyama
Priority date: 1998-06-22
Filing date: 1999-06-17
Publication date: 1999-12-29

Abstract

An image correcting method and a digital image inputting device such as a digital camera or a video camera, wherein various aberrations of an image forming optical system and errors in the color conversion function of an image sensor are corrected for respective devices, by digital techniques, and a practically high-enough image quality is realized by means of a cheap, simple image forming optical system. An image representing a predetermined proofing figure is captured. Image correcting information for converting the image into an image to be present is derived for respective digital image inputting devices and stored in a device or system. After normal imaging, the image is converted to a correct one by using the image correcting information. The image correcting method comprises Fourier-transforming the picked-up image by using image correcting information for correcting the transfer function of an optical system to a design transfer function in a spatial frequency region, correcting the transformed image in a spatial frequency region, and forming a corrected image by inverse Fourier transform.

Description

Description Image correction method and image input device

Technical field

The present invention relates to a method of correcting an image obtained from a digital image input device such as a digital camera or a video camera which converts an image into an electric signal and records the image, and in particular, various distortions of an imaging optical system and a function of converting an image sensor. The present invention relates to an image correction method and an image input device capable of individually correcting defects and the like of the above. Background art

In a conventional camera, an image of a subject is formed by an imaging optical system, and in a film camera, the image is recorded by converting it into an electric signal for each pixel by an image sensor in a digital camera due to a chemical change. Normally, an image sensor using a CCD is used for the image sensor. However, in the case of color, a color filter is required for the structure of the image sensor, and the primary colors of pixels, such as red, green, and blue, are viewed from the imaging optical system. However, the conditions are not always the same, and severe conditions regarding the aberration are imposed on the imaging optical system. Of course, from the viewpoint of the fidelity of the image to the subject, there are various aberrations including chromatic aberration as a common problem in the image forming optical system. Both have various problems.

It is difficult to completely solve all of these problems, and ultimately compromises must be made to satisfy the average. Nevertheless, a great deal of effort has been made in the materials and design of the lenses that make up the imaging optical system. It is devastated and requires delicate adjustment work during production.

Therefore, an object of the present invention is to provide a system that can be called a computer lens, which realizes an image faithful to an object by a digital technology even if the adjustment operation after manufacturing is not sufficient, even with a cheap and simple imaging optical system. The purpose is to propose a digital image capture method and image input device that can be realized. Disclosure of the invention

The basic concept of the present invention is to capture the image of the graphic for calibration into a digital image input device such as a digital camera, video camera, etc., separately derive and store image correction information for a desired image for each device, and shoot the image. Each time an image is captured, or afterwards, a predetermined operation is performed based on the image correction information to correct and complete the image.

A system for performing image correction can be performed by a control unit and a calibration program built into a digital image input device, or by a dedicated device or a method in which a personal computer has image correction software. is there.

The image correction method has a calibration mode and a photographing 'correction mode. In the calibration mode, a known calibration figure is photographed, the captured image is subjected to Fourier transform, and a desired image is obtained for the calibration figure. The image correction information is formed in the spatial frequency domain except for the Fourier transform. In the shooting and correction mode, the captured image is Fourier-transformed, multiplied by the image correction information in the spatial frequency domain, and then the corrected image is obtained by Fourier inverse transform. This method requires a considerable amount of time for calculation processing, so it has the advantage of being able to compensate for high computational power and reduced power resolution that requires the use of a control unit.

In addition, not only image distortion, but also chromatic aberration of the imaging optical system and error of color change capability of the image sensor can be targeted for correction. In other words, image correction information is generated for each primary color of the image sensor, and correction is performed for each primary color, so that the chromatic aberration in the imaging optical system or the environmental problem for the image between the primary colors due to a structural problem of the image sensor. Disadvantages such as differences in Alternatively, a correction image that can correct the intensity of each primary color of a pixel is generated by including a pattern in a calibration image that can calibrate parameters that affect color reproducibility such as lightness and saturation of color, and generate correction information. The image after shooting can be corrected. BRIEF DESCRIPTION OF THE FIGURES

Fig. 1 is a diagram for explaining the basic concept of the present invention, which shows a model of an object, an optical system, an image, and the like, and also shows a coordinate system.

Figure 2 shows the intensity of the corresponding image when the optical system is distorted using a point light source as the object figure. Show cloth.

FIG. 3 shows the procedure of the image correction method according to the first embodiment.

FIG. 4 shows a calibration figure having a point light source for each section used in the second embodiment.

Figure 5 shows an example of an image corresponding to a calibration figure having a point light source for each section.

Figure 6 shows the procedure for the calibration mode of the second embodiment.

Fig. 7 shows the procedure of the shooting and correction mode of the second embodiment.

Figure 8 shows an example of image segmentation in the shooting and correction mode.

Figure 9 shows a diagram for explaining the concept of the segmented window function.

FIG. 10 is an external view of a digital camera according to the third embodiment.

FIG. 11 shows a functional block diagram of a digital camera according to the third embodiment. BEST MODE FOR CARRYING OUT THE INVENTION

The present invention relates to a digital image input device having at least an imaging optical system and an image sensor, which converts an image of a subject into a digital electric signal and captures the image. We propose a method for compensating for. The basic concept of the present invention will be described with reference to FIGS. 1 and 2, and then the image correction method and the image input device of the present invention will be described with reference to embodiments. FIG. 1 is a diagram for explaining the basic concept of the present invention, and shows a model of the relationship among a subject, an imaging optical system, an image, and the like. In the figure, the object is indicated by number 11, the lens that is the imaging optical system is indicated by number 12, and the image input to the image sensor 1 is indicated by number 13. The coordinates of the plane where the object is located are (X, y), the coordinates of the plane with the image are (χ ', y'). The amplitude distribution u (χ ', y') of the image can be approximated from the subject's amplitude distribution g (x, y) in the form of a convolution integral as shown in Eq. (1). The function h (x, y) shown in Eq. (1) is a transfer function that represents the characteristics of the optical system, and includes information such as various aberrations and chromatic aberrations of the optical system. The transfer function of this optical system can be easily obtained by placing a point light source represented by δ (X) δ (y) as the subject, since h (X, y) becomes equal to the amplitude distribution of the image. Here, δ (X) and δ (y) indicate delta functions. The meaning of this transfer function will be described in more detail with reference to FIG. FIG. 2 (a) shows a subject, and in this case, an example using a point light source is shown at number 21. The torsional width distribution is an amplitude distribution without spread as shown in No. 24 when viewed along the line of No. 22 and No. 25 when viewed along the line of No. 23. However, the image formed on the image sensor via the imaging optical system has a spread and a bias as shown by 31 and 32 in Fig. 2 (b). No. 32 is a contour plot showing the torsion width distribution of the image in a visual manner. In this figure, the amplitude distribution of the image along the line of No. 33 is No. 35, and As shown in No. 36, the amplitude distribution along the line 34 spreads out from the original amplitude distribution of the point light source, and becomes blurred or biased. As explained in relation to equation (1), the object is a point light source, and the distributions indicated by numbers 31 and 32 indicate the transfer functions including various imperfections such as distortion and chromatic aberration of the actual optical system. u (x, y), g (x, y), h (x, y) The Fourier transform form of each function is F [u (x, y)], F [g (x, y)], F [ h (x, y)], the Fourier transform expression of the convolution integral of Eq. (1) is expressed by Eq. (2). Here, X and Y are spatial frequencies, and the definition of the Fourier transform is shown in Eq. (3).

(1) u *, y ') = I I g (x, y) h (x,-x, y-y dxdy

(2) F [u (x ', y')] = F [g (x, y)] F [h (x, y)]

(3)

F [u (x ', y')] u (χ ', y') exp [-i2n (+ '+ Yy')] dx 'dy' However, the signal detected by the image sensor is not the amplitude, but the energy Therefore, we consider the time average of the square of the amplitude, that is, the relational expression in the intensity distribution. Assuming that the subject figure and the image have a time term, they are denoted by g (X, y, t) and u (χ ', y', t), respectively. Then, the intensity distribution I (χ ', y') of the image is expressed by Eq. (4).

Here, u * (χ ', y,, t) is a function of the conjugate of u (χ', y ', t), and> indicates the time average.

(4) I (x'.y ')-(u (x', y ', t) u * (x', y ', t)>

= <g (x, y, t) h (x -x, y -y) dxdy g * (x, y, t) h * -x, y 'y) dxdy>

Since the target light is generally spatially incoherent, Eq. (4) is expressed by the convolution integral as shown in Eq. (5), and the Fourier transform form is Eq. (6).

(5) I (χ ', y') = | g (x, y) | ^J | h (x -x, y -y) | ^J dxdy

(6) F [I ( _X ', y')] = F [j ( _X , y) 3F [| h (x, y) I ² ]

Here, J (x, y) indicates the intensity distribution of the calibration figure and is given by Eq. (7).

(7) J (x, y) = <g (x, y, t) g * (x, y, t)>

On the other hand, the transfer function of the desired optical system is h. Expressed as (x, y), the Fourier transform form of the image intensity distribution I 0 (x ′, y ′) in that case is given by Eq. (8). From Equations (6) and (8), the Fourier transform form of the image for which it is desirable to eliminate the intensity distribution of the unknown object figure is expressed by Equation (9).

(8) F [I. (x ', y'>] = F [j (x, y)] F [| ho (x, y) l ² ]

(9). ()] = ¾ ^)] Therefore, if the function shown on the right side of Equation (10) is used as image correction information and stored and stored for each image input device, it is possible to correct the image after photographing to a desired image by the optical system. In other words, the Fourier transform form of the image during normal shooting corresponds to F [I (χ ', y')]. Therefore, after multiplying the image correction information according to Eq. (9), the desired result is obtained by inverse Fourier transform. To obtain an image equivalent to using an optical system. However, the equation for the inverse Fourier transform is shown in (11).

_{(1 0)} Fi x'.y ')! F [| ho (x, y) l ^J ]

F [l (x ', y)] F [| h (x, y) | ² ]

(1 1> (x ', y) = F [Ι. ( _Χ ', y ')] exp [i2n (Xx' + Y) JdXdY Here, the expression (10) is given for each image input device. It is necessary to obtain and store such image correction information in advance, but it can be obtained by photographing a calibration figure as is clear in the process of deriving equation (10) from equations (6) and (8). The calculation can be derived by subtracting the Fourier transform form of the image obtained using the desired optical system given as the design value in the Fourier transform form of the intensity distribution of the image. In Equation (6), the image is equivalent to I h (x, y), which is the autocorrelation function of the transfer function h (x, y). Since they exist discretely within a limited range, the Fourier transform and inverse transform in each of the above equations are exclusively performed by the FFT algorithm. (9), (10) The product or division in the spatial frequency domain in the equation is performed by matrix operation, and since it is well known, detailed description of this point is omitted. The concept of the correction was explained, but the procedure of the image correction method is shown as a first embodiment in Fig. 3. The first embodiment has a calibration mode and a shooting and correction mode, each of which is implemented in the following procedure. Is done.

First, in the calibration mode, the conditions for shooting, such as the distance and aperture, are set in step (1), and the image of the calibration figure is captured in step (2) and used as the first image. In step (3), a desired image corresponding to the graphic for calibration is read out and set as a second image. In step (4) One of the primary colors of the di-sensor is selected and the image is processed between the components of this primary color. In step (5), the Fourier transform of the second image is divided by the Fourier transform of the first image to obtain image correction information. In step (6), the selection of primary colors is changed, and step (5) is repeated to obtain image correction information for each primary color. Further, in step (7), the aperture is changed, and steps (1) and (6) are repeated to obtain image correction information corresponding to a plurality of aperture values.

In the "photographing" correction mode, the image captured in step (8) is fetched, and in step (9), the primary color of the image sensor is selected, and thereafter, the image of this primary color component is processed. In step (10), image correction information corresponding to the aperture at the time of photographing the image to be corrected is read. When there is no image correction information of a condition that matches the aperture, the image correction information is obtained by interpolating from the preceding and following image correction information. In step (11), the Fourier transform form of the image in step (8) is multiplied by image correction information, and in step (12), inverse Fourier transform is performed to obtain a corrected image for one primary color. In step (13), the primary colors are changed, and steps (10)-(12) are repeated to obtain a corrected image for each primary color to complete the image correction. Image correction information is obtained for each primary color of the image sensor, and correction is performed for each primary color component because the optical system does not always have the same conditions between the primary colors. Naturally, care must be taken into account in the design, but there are quite a few adjustment steps, and if image correction is assumed as in the first embodiment, the cost increase factors can be reduced considerably. . In addition, if there is another zoom lens as an element that changes the state of image distortion, it must also be considered. In the procedure of the first embodiment, image correction information should be obtained for each of the magnifications by the zoom lens as well as the aperture. In photographing, it is, of course, generally desirable to faithfully image the subject. Ideally, a point light source approximated by the delta function would be an image represented by the delta function, and in that sense, the ideal transfer function would be the delta function. Theoretically, since the Fourier transform form of the delta function is a constant, the real image for the point light source is Fourier transformed and the ideal image correction information is the reciprocal thereof. However, under these conditions, overcorrection may actually occur and another distortion may occur. That is, the mathematical Fourier transform domain is infinite However, the aperture of the optical system is finite, and the distance between pixels of the image sensor for taking in digital images is also finite. From this point as well, the frequency components included in the Fourier-transformed image have both low-frequency and high-frequency limits, and it is impossible to amplify or compress even frequency components that do not exist. There is. The target design transfer function in the process of image correction is set in consideration of the quality of the imaging optical system used, the resolution of the image sensor, and the like. The point of using the calibration system is to accurately match the conditions such as the position and the size of the figure between the captured image and the image that should be a desirable optical system at the time of calibration. A point figure was used in the calibration figure described above. In addition to the central point figure, a similar point figure was also placed at a distant position, and in the real space area before image correction information was calculated in the spatial frequency domain. The problem can be easily solved by normalizing the position, size, etc. of the points based on the centers of the point figures. However, moving and expanding / contracting is performed on the side of the image that should be a desirable optical system, and only the point figure near the center of the area is used to calculate the image correction information, simplifying the process by excluding others. . In the first embodiment, in which image correction is performed using image correction information in the frequency domain, image resolution can be compensated to some extent and the field of application is wide, but on the other hand, conditions that can be applied in practice are: There is a tough side. That is, the theoretical basis is based on the validity of the approximation of Eq. (1). Force The transfer function actually changes depending on where the subject is located with respect to the optical axis. If the aperture of the imaging optical system has a sufficiently small aperture and the light passing through the imaging optical system is only near the central axis, it is considered correct in a considerable range if the aperture is large. The image has large distortion. For example, if a plurality of point light sources 43 are dispersed throughout as shown in Fig. 4, the image can be exaggerated and expressed in different forms as indicated by numbers 53, 54, 55 in Fig. 5. Becomes Also, if the optical system includes a zoom lens, the transfer function will be different at each magnification. The image correction method shown in the second embodiment provides a practical solution in such practical application, and will be described below with reference to FIGS. Fig. 4 shows a calibration graphic 41 used in the second embodiment. The whole is divided into a plurality of sections 42, and a point light source 43 is arranged in each section. FIG. 5 shows an image 51 in which the graphic 41 for calibration is captured, and is assumed to be divided into a plurality of sections indicated by reference numeral 52. The image of the point light source 43 exists in those sections, but the shape differs between the center and the periphery as shown by numbers 53, 54, and 55 depending on the location of the section. The concept of the image correction method described in the second embodiment is based on this situation, and the screen is divided into a plurality of sections, image correction information in the frequency domain is obtained for each section, and the image is corrected for each section. , Images are combined and completed. FIG. 6 illustrates the procedure of the calibration mode in the image correction method in the second embodiment, and FIG. 7 illustrates the procedure of the imaging / correction mode in FIG. In the calibration mode shown in Fig. 6, the screen is divided into multiple sections in step (1). In step (2), the photographing conditions such as the distance from the calibration figure and the aperture are set. In step (3), as shown in Fig. 4, an image of a calibration figure in which point light sources are arranged for each section is captured. In step (4), the primary color of the image sensor is selected, and the image of this primary color component is processed thereafter. In step (5), a Fourier transform of the image of the point light source is performed for each section. In step (6), a Fourier transform of a correct image that should have a point light source is performed. This is common to all sections, and the Fourier transform type may be obtained in advance and stored in memory. However, normalization of the position, size, etc. requires the same processing as described in the first embodiment. In step (7), the Fourier transform form of the image obtained in step (6) is divided by the Fourier transform form obtained in step (5) to obtain divided image correction information in the frequency domain. In step (8), select another primary color and repeat steps (5)-(7). In step (9), the aperture is changed, and steps (3)-(8) are repeated to obtain the segmented image correction information in the frequency domain for multiple apertures. FIG. 7 shows the procedure in the photographing'correction mode. Take an image in step (10). In step (11), the primary colors of the image sensor are selected, and processing is performed between images of the components of these primary colors. Select the section in the screen in step (12). In step (13), the image in the section is multiplied by the window function to perform Fourier transform. In step (14), the divided image information in the frequency range suitable for the aperture at the time of shooting is read. If there is no section image correction information suitable for the aperture at the time of shooting Interpolate or extrapolate from the preceding and following sectioned image correction information. In step (15), the Fourier transform obtained in step (13) is multiplied by the segmented image correction information in the frequency domain read in step (14), and a corrected image for each segment is obtained by inverse Fourier transform. In step (16), a new section is set so that adjacent parts overlap, and steps (13) to (15) are repeated to scan the entire area of the screen to complete the corrected image. In step (17), the selection of the primary color is changed, and steps (12) to (16) are repeated. Here, the sections and calculations in steps (12) and (13) and the corrected image for each section obtained in step (15) will be described with reference to FIGS. In FIG. 8, the number 81 indicates the section corresponding to the section 52 in FIG. 5, but the section shown in step (12) in FIG. 7 indicates the larger sections 82, 83, 84, etc. The window function to be multiplied in step (13) is a weight function such that it becomes zero at the boundary of the section and 1.0 at the center. This is the curve with the number 91 shown in Fig. 9 in comparison with the section 82, which is a two-dimensionally distributed surface of (x, y), but Fig. 9 shows a cross section. Since the Fourier transform targets a periodic function, the boundary conditions are adjusted in this way. Therefore, even if the image is multiplied by the window function and the image corrected in step (15) is obtained through the inverse Fourier transform, a correct result is not obtained near the boundary of the segment. Only the portion at the center 85 of is corrected. In step (16), the adjacent sections overlap so that the corrected image is connected. For example, the section is selected and moved, such as section 83 next to section 82, or section 84 above and below, and the entire screen is sequentially moved. Scan. One of the important objectives of the present invention is to use a low-cost but inexpensive imaging optical system and improve the image quality to the level of high-end products by digital processing. From that point of view, rather, digital image input equipment with poor quality imaging optics, such as the one shown in Figure 5 with a point light source image, should be the target. The second embodiment is based on the assumption that the transfer function of the imaging optical system continuously changes while allowing the transfer function to change depending on the location. The image correction information obtained in step (1) is obtained, and the images are corrected in a piecewise manner and combined. In the first embodiment, the calculation of the Fourier transform and the inverse transform are performed on the entire screen. Only However, in the second embodiment, the entire screen is divided into multiple sections, and the Fourier transform and inverse transform are performed for each section. According to recent trends, image sensors with more than 100,000 pixels are becoming more common. However, rather than performing a Fourier transform or inverse transform on 100,000 pixels at a time, they operate in a piecewise manner. It is much easier to do this, and it is also easier to use dedicated hardware for the arithmetic unit for speeding up. In the first and second embodiments, the Fourier transform and the inverse transform are indispensable on the premise of correction in the frequency domain. Although the formulas in the above description are expressed as continuous quantities, the pixels on the image sensor exist discretely, and are converted to a discrete calculation form in accordance with this pixel distribution. Since they are well known, they have not been described especially with reference to discrete expressions. Naturally, the Fourier transform is performed at high speed using a discrete model, so the Fourier transform and the inverse Fourier transform are performed using the FFT. However, even if the FFT is used for the Fourier transform, the number of pixels is large. In short, image correction performed in parallel with shooting may not be at a practical level at the current stage of computation unit. However, as in the second embodiment, if the division can be determined and the FFT operation can be made into dedicated hardware, the time can be significantly reduced. The image correction method according to the first and second embodiments of the present invention has been described. This image correction method is a method of correcting an image by a digital image input device based on image correction information unique to the digital image input device to obtain a correct image. It can be collected and generated, and the necessary programs can be installed in a personal computer, or it can be made by dedicated equipment. Of course, it can also be installed in individual devices such as digital cameras. The third embodiment will be described with reference to FIGS. 10 and 11 using a digital camera as an example. FIG. 10 shows the appearance of a digital camera 100 according to a third embodiment of the present invention. In the figure, number 101 is the lens which is the imaging optical system, number 102 is the viewfinder and the liquid crystal display for displaying the image after shooting, number 103 is the shutter button, Number one Reference numeral 04 denotes a mode selector for switching various functions, reference numeral 105 denotes a power switch, and reference numeral 106 denotes a connector for connection to an external device. FIG. 11 is a functional block diagram of a digital camera 100 according to a third embodiment of the present invention. The image of the subject is formed on the image sensor 111 by the lens 101 which is an image forming optical system, and is activated by pressing the shutter button 103, and the control unit 112 is controlled by the image sensor 111. To convert the image into an electric signal and store it in the memory 114. The numbers 1 13 indicate the control memory in the control unit 112, and the numbers 115 indicate the batteries. The mode can be switched between the normal shooting mode and the calibration mode by the instruction of the mode selector 104. In the calibration mode, a predetermined calibration figure 1 16 is used as the subject, and the distance or brightness between the calibration figure 1 16 and the lens 101, as well as various conditions such as aperture and shutter speed, are determined. After setting, the control unit 112 loads the image of the calibration figure 111 into the memory 114. After that, the control unit 112 divides the screen according to a predetermined division method, performs a Fourier transform for each section, and stores a part of the control memory 113 from the design-required optical system. Based on the Fourier transform type of the image, the divided image correction information is calculated and stored in the control memory 113. These processes and operations are executed by the control unit 112 instructed by a program following the procedure shown in the second embodiment. This processing program is stored in the control memory 113. During normal shooting, the image captured by the lens 101 and the image sensor 102 is recorded in the memory 114, and the image correction information stored in the control memory 113 is stored in the memory 114. At the same time, the control unit 112 executes the image correction in accordance with the instruction of the program stored in the control memory 111, and stores the image in the memory 114. However, even if the FFT calculation method is used for the Fourier transform, the Fourier transform and the inverse transform require a considerable amount of time, so the image is stored together with the shooting conditions without performing image correction immediately after normal shooting. It would be more realistic to store the data in 14 and perform the image correction separately according to the instruction from the mode selector 104. Alternatively, in the calibration mode, only the image of the calibration figure is captured, and in the shooting mode, only the shooting conditions and the image are captured. It is also possible to collectively collect images later using a computer or dedicated machine. Although the embodiment of the present invention has not been particularly described with reference to the drawings, improving the reproducibility of color information is also included in the present invention. Color reproduction is also a major problem in color images, but there are chromatic aberrations in the imaging optics, and there are also differences in sensitivity to color in image sensors and, in some cases, at each pixel level. It is a social loss to increase all costs due to all these errors in the manufacturing process. However, according to the present invention, it is possible to solve the same problem by improving the distortion. In other words, the chromatic aberration of the imaging optical system can be completely corrected by performing correction with the image correction information for each primary color of the image sensor. Furthermore, by including correction patterns for the color reproducibility in the calibration pattern used in the calibration mode and deriving correction information for each pixel, it is possible to resolve sensitivity differences depending on the location of the image sensor. is there. For example, a pattern with different parameters related to color reproducibility, lightness, saturation, etc. is appropriately placed on the screen and the image is captured into the digital camera. Since the color attributes in each pattern in the image are known in advance, how to read from the obtained image is used as image correction information. In this way, the image correction information is determined, and when the image is corrected, the output is read for each pixel in accordance with the image correction information and the correction is performed. In this process, the color conversion function of the image sensor is corrected, including the chromatic aberration in the imaging optical system, so that the fidelity of color can be improved. As described above with reference to the embodiment, the image correction method is described in the first and second embodiments, and the image correction method is programmed and stored in the digital image input device in the third embodiment. The example of executing the calibration and the image correction has been described. According to the present invention, according to the present invention, it is possible to provide a complete system with all the programs or data necessary for calibration in an image input device, or to use a program necessary for a system such as a general-purpose personal computer. It is also possible to correct images after shooting, and those programs are the subject of the present invention. In the latter case, it will be important to prevent complications during image correction by combining the captured image with a unique identification number that includes the serial number of the digital image input device. Industrial applicability

According to the present invention, as described in the embodiment with a digital camera, in a digital image input device, an image capable of correcting imperfections of an image forming optical system or an image signal conversion system of each device. It has been described that the correction information is stored, and the correct or higher quality image is obtained by using the image correction information every time the image is obtained or collectively afterwards, and can be easily realized by the present invention. . As described above, the present invention has shown that a digital image having a practical level of quality can be obtained by digital technology even if an inexpensive and low-quality imaging optical system or image sensor is used.

The present invention can provide a method of realizing low cost as a system including not only completed devices but also manufacturing processes. In other words, if any adjustment is required between the chromatic aberration or the primary colors of the image sensor in the manufacturing process, the calibration figure is read and then left to the calibration system in the device to omit the adjustment in the manufacturing process. It is possible to do so and there are ways to reduce costs.

In general, equipment and materials often cannot maintain their initial performance due to deterioration or misalignment of adjustment points over time after production. In the present invention, calibration is performed after the production of digital image input devices such as digital cameras and video cameras, and the quality performance is demonstrated with the help of digital processing technology. Should be recognized. The present invention is also effective in this respect, and a calibration program can be built in the apparatus, or re-calibration can be performed by a specialty store or a manufacturer having a calibration control apparatus to maintain design performance. . This is also one of the important objects of the present invention.

Claims

The scope of the claims

1. In an image input device having at least an optical system and an image sensor, an image corresponding to a known calibration figure is obtained, and image correction information unique to the optical system is formed in the spatial frequency domain. An image correction method and an image input device characterized by correcting an output image of an image input device include at least the following steps.

(1) Capture a known calibration figure as the first image and perform Fourier transform.

(2) Fourier transform the correct image corresponding to the graphic for calibration as the second image.

(3) Divide the Fourier transform form of the second image by the Fourier transform form of the first image to form and store image correction information in the frequency domain.

(4) Fourier transform the output image of the image input device, multiply by the image correction information formed in step (3), and perform inverse Fourier transform to obtain a corrected image.

2. The image correction method and the image input device according to claim 1, wherein the calibration graphic in steps (1) and (2) is a point graphic or a point light source. And image input devices.

3. In the surface image correction method and the image input device described in claim 1, image correction information is obtained for each primary color of the image sensor, and a corrected image is obtained by correcting each primary color in step (4). And an image input device.

4. In the image correction method and the image input device according to claim 1, each image correction is performed by steps (1), (2), and (3) for each of a plurality of imaging conditions such as an aperture and a focal length. The information group is derived and stored, and optimal image correction information is generated by interpolation or extrapolation of the image correction information group according to imaging conditions, and is used for image correction in step (4). Image correction method and image input device

5. In the image correction method and the image input device according to claims 1, 3, and 4, the target image is divided into a plurality of sections, and the image is taken from a known calibration figure. Output image Is subjected to Fourier transform for each section to form unique segmented image correction information in the spatial frequency domain, and the output image of the digital image input device is corrected for each section in the spatial frequency domain using the segmented image correction information. An image correction method and an image input device characterized by combining are provided with the following steps.

(1) Divide the target screen into multiple sections.

(2) Capture a known calibration figure as the first image and perform Fourier transform for each section.

(3) Fourier transform is performed for each section as a second image, which should be a correct image corresponding to the calibration figure.

(4) Dividing the Fourier transform type for each section of the second image by the Fourier transform type for each corresponding section of the first image to form and store frequency domain sectioned image correction information.

(5) Fourier-transform the output image of the digital image input device for each section, multiply by the frequency-domain section image correction information formed in step (4), and then perform Fourier inverse transform for each section to correct for each section. Image.

(6) The corrected images for each section obtained in step (5) are connected and combined to obtain a corrected image.

6. The image correction method and the image input device according to claim 5, wherein the calibration graphic in steps (2) and (3) is a point graphic or a point light source arranged in a plurality of sections. Characteristic image correction method and image input device

7. The image correction method and the image input device according to claim 5, wherein the sections divided in step (1) overlap each other for each adjacent section. And image input equipment