Image processing device, printing system, and image processing method

Abstract

To provide a technique capable of acquiring a converted image in which thin lines are appropriately connected when a conversion magnification of a resolution is different according to directions.SOLUTION: An image processing device includes: a blur processing unit capable of generating a blur image by applying a two-dimensional gaussian filter having an S1xS2 filter size to an input image; and a resolution conversion unit capable of generating, from the blur image, a converted image in which a resolution is converted with a first conversion magnification R1 of a first direction and a second conversion magnification R2 of a second direction. When the blur processing unit employs S1>S2, R2>R1 is realized. When the blur processing unit employs S2>S1, R1>R2 is realized. In this case, a standard deviation σ of the gaussian filter in the first direction and a standard deviation σ of the gaussian filter in the second direction are respectively set to standard deviations σ1 and σ2. When S2 is different from S1, the blur processing unit applies the gaussian filter having a filer coefficient in which σ2=σ1 to the input image.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a technique for converting the resolution of an image.

Background Art

[0002] In order to convert the resolution of an image having a plurality of pixels arranged in two mutually orthogonal directions, the number of pixels of the image is converted by an interpolation operation such as the bicubic method. Converting to increase the resolution of the image increases the number of pixels of the image, and converting to decrease the resolution of the image decreases the number of pixels of the image. Here, let the two mutually orthogonal directions be the X direction and the Y direction. In particular, when the resolution of the image becomes low, among the plurality of pixels included in the image, pixels that are not used for resolution conversion or pixels with a small weight during interpolation operation occur. Therefore, a blurring process of applying a Gaussian filter to the image before resolution conversion is performed.

[0003] The printed matter inspection apparatus disclosed in Patent Document 1 performs smoothing on the RIP (Raster Image Processor) image before resolution conversion using a Gaussian filter. The filter size of the Gaussian filter used is the same size in the X direction and the Y direction, such as 3×3. The smoothing coefficient σ of the Gaussian filter, that is, the standard deviation, becomes larger as the printing line number of the printed matter is lower.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] When the resolution conversion ratio differs between the X and Y directions, and an image containing fine lines undergoes resolution conversion, the thickness of the fine lines in the converted image may change inappropriately or the fine lines may be broken depending on the conversion ratio in each direction. For example, if a Gaussian filter with the same filter size is applied to an image in both the X and Y directions, and the smoothing coefficient σ increases as the resolution decreases, then if the conversion ratio in the Y direction is smaller than that in the X direction, fine lines along the Y direction will become thinner and thicker. Fine lines in directions closer to the Y direction than the X direction may be broken. Therefore, it is desirable to appropriately convert fine lines while obtaining a suitable blurring effect, even when the resolution conversion ratio differs between the X and Y directions. [Means for solving the problem]

[0006] The present invention is an image processing apparatus capable of performing image processing to acquire a converted image whose resolution has been converted based on an input image having a plurality of pixels arranged in a first direction and a second direction intersecting the first direction, A blurring processing unit capable of generating a blurred image by applying a two-dimensional Gaussian filter, which has a first filter size in the first direction and a second filter size in the second direction, to the input image, The system comprises a resolution conversion unit capable of generating a converted image from the blurred image, wherein the resolution has been converted by a first conversion ratio, which is the resolution conversion ratio in the first direction, and a second conversion ratio, which is the resolution conversion ratio in the second direction. If the blurring processing unit makes the first filter size larger than the second filter size, the second conversion magnification is larger than the first conversion magnification. If the blurring processing unit makes the second filter size larger than the first filter size, the first conversion magnification is larger than the second conversion magnification. Let the standard deviation σ of the Gaussian filter in the first direction be denoted as standard deviation σ1. Let the standard deviation σ of the Gaussian filter in the second direction be the standard deviation σ2. The blurring processing unit has an embodiment in which, when the second filter size is different from the first filter size, a Gaussian filter having filter coefficients in which the standard deviation σ2 matches the standard deviation σ1 is applied to the input image.

[0007] Furthermore, the printing system of the present invention is The image processing apparatus and, A print head having multiple nozzles capable of dispensing liquid onto a medium, The embodiment includes a control unit that controls the discharge of the liquid from the print head so that a printed image based on the converted image is formed on the medium.

[0008] Furthermore, the present invention is an image processing method that acquires a converted image whose resolution has been converted based on an input image having a plurality of pixels arranged in a first direction and a second direction intersecting the first direction, A blurring process that generates a blurred image by applying a two-dimensional Gaussian filter, which has a first filter size in the first direction and a second filter size in the second direction, to the input image; The process includes a resolution conversion step of generating a converted image from the blurred image, in which the resolution has been converted by a first conversion ratio, which is the resolution conversion ratio in the first direction, and a second conversion ratio, which is the resolution conversion ratio in the second direction. In the blurring process described above, if the first filter size is made larger than the second filter size, the second conversion magnification is made larger than the first conversion magnification. In the blurring process described above, if the second filter size is made larger than the first filter size, the first conversion magnification is made larger than the second conversion magnification. Let the standard deviation σ of the Gaussian filter in the first direction be denoted as standard deviation σ1. Let the standard deviation σ of the Gaussian filter in the second direction be the standard deviation σ2. In the blurring process, if the second filter size is different from the first filter size, the Gaussian filter having filter coefficients such that the standard deviation σ2 matches the standard deviation σ1 is applied to the input image. [Brief explanation of the drawing]

[0009] [Figure 1] A schematic block diagram showing an example configuration of a printing system including an image processing device. [Figure 2] This diagram schematically illustrates an example of generating a transformed image where a horizontal Gaussian filter is used for blurring, and then the vertical transformation ratio is greater than the horizontal transformation ratio. [Figure 3] This diagram schematically illustrates an example of generating a transformed image where a vertical Gaussian filter is used for blurring, and then the horizontal transformation ratio is greater than the vertical transformation ratio. [Figure 4] A schematic diagram illustrating an example of the relationship between the resolution conversion ratio and the filter size of a Gaussian filter. [Figure 5] This diagram schematically shows examples of images when a blurring process using a horizontal Gaussian filter is performed, and then a transformed image is generated in which the vertical transformation ratio is greater than the horizontal transformation ratio. [Figure 6] A schematic diagram showing an example of diagonal lines after resolution conversion. [Figure 7] This diagram schematically shows an example of a converted image obtained from an input image containing vertical thin lines by changing the resolution in the Y direction. [Figure 8] A schematic diagram showing examples of standard deviations σi associated with each printing condition. [Figure 9] A flowchart schematically illustrating an example of print control processing. [Figure 10] A schematic diagram illustrating an example of a user interface screen. [Figure 11] A flowchart schematically illustrating an example of the blurring image generation process. [Figure 12] A schematic diagram showing an example of filter coefficients for a Gaussian filter. [Figure 13]A diagram schematically showing a comparative example of a converted image obtained from an input image including vertical thin lines by changing the resolution in the Y direction. [Embodiments for Carrying Out the Invention]

[0010] Hereinafter, embodiments of the present invention will be described. Of course, the following embodiments are merely illustrative of the present invention, and not all of the features shown in the embodiments are necessarily essential to the solution means of the invention.

[0011] (1) Outline of the technology included in the present invention: First, the outline of the technology included in the present invention will be described with reference to the examples shown in FIGS. 1 to 13. Note that the figures in the present application are diagrams schematically showing examples, and the magnification ratios in each direction shown in these figures may be different, and the figures may not be consistent. Of course, each element of this technology is not limited to the specific examples indicated by the reference numerals. In the "outline of the technology included in the present invention", the content in parentheses means a supplementary explanation of the immediately preceding word.

[0012] [Aspect 1] An image processing apparatus U0 according to one aspect of this technology is an image processing apparatus U0 capable of performing image processing to acquire a converted image IM2 whose resolution has been converted based on an input image IM1 having a plurality of pixels PX1 arranged in a first direction D1 and a second direction D2 intersecting the first direction D1, as illustrated in Figures 2, 3, etc. The image processing apparatus U0 comprises a blurring processing unit U1 and a resolution conversion unit U2. The blurring processing unit U1 can generate a blurred image IM11 by applying a two-dimensional Gaussian filter F0 having a first filter size S1 in the first direction D1 and a second filter size S2 in the second direction D2 to the input image IM1. The resolution conversion unit U2 can generate the converted image IM2 from the blurred image IM11, in which the resolution has been converted by a first conversion ratio R1, which is the resolution conversion ratio in the first direction D1, and a second conversion ratio R2, which is the resolution conversion ratio in the second direction D2. As illustrated in Figure 2, when the blurring processing unit U1 makes the first filter size S1 larger than the second filter size S2, the second conversion magnification R2 is larger than the first conversion magnification R1. As illustrated in Figure 3, when the blurring processing unit U1 makes the second filter size S2 larger than the first filter size S1, the first conversion magnification R1 is larger than the second conversion magnification R2. Here, the standard deviation σ of the Gaussian filter F0 in the first direction D1 is denoted as standard deviation σ1, and the standard deviation σ of the Gaussian filter F0 in the second direction D2 is denoted as standard deviation σ2. When the second filter size S2 is different from the first filter size S1, the blurring processing unit U1 applies the Gaussian filter F0 having a filter coefficient KE in which the standard deviation σ2 matches the standard deviation σ1 to the input image IM1.

[0013] The blurring processing unit U1 in the above embodiment can apply a Gaussian filter F0 in which the second filter size S2 is different from the first filter size S1 to the input image IM1. On the other hand, even if the second filter size S2 is different from the first filter size S1, the standard deviation σ2 of the Gaussian filter F0 in the second direction D2 is the same as the standard deviation σ1 of the Gaussian filter F0 in the first direction D1.

[0014] As illustrated in Figure 2, when the second conversion magnification R2 of the resolution in the second direction D2 is greater than the first conversion magnification R1 of the resolution in the first direction D1, and a Gaussian filter F0 with a first filter size S1 greater than the second filter size S2 is applied to the input image IM1, the following effect is obtained. Because the blur range in the first direction D1 is wider than the blur range in the second direction D2, the connection of thin lines L0 in the direction closer to the first direction D1 than to the second direction D2 is ensured. Also, because the blur range in the second direction D2 is narrower than the blur range in the first direction D1, the blur in the second direction D2 is not excessive. Furthermore, in the Gaussian filter F0, the standard deviation σ2 of the second direction D2 matches the standard deviation σ1 of the first direction D1, so the blur effect in the first direction D1 and the second direction D2 is appropriate, as is the blur effect in the diagonal direction between the first direction D1 and the second direction D2.

[0015] As illustrated in Figure 3, when the first conversion magnification R1 of the resolution in the first direction D1 is greater than the second conversion magnification R2 of the resolution in the second direction D2, and a Gaussian filter F0 with a second filter size S2 greater than the first filter size S1 is applied to the input image IM1, the following effect is obtained. Because the blur range in the second direction D2 is wider than the blur range in the first direction D1, the connection of the thin line L0 in the direction closer to the second direction D2 than to the first direction D1 is ensured, as illustrated in Figure 6. Also, because the blur range in the first direction D1 is narrower than the blur range in the second direction D2, the blur in the first direction D1 is not excessive. Furthermore, since the standard deviation σ1 of the first direction D1 in the Gaussian filter F0 matches the standard deviation σ2 of the second direction D2, the blur effect in the second direction D2 and the first direction D1 is appropriate, as is the blur effect in the diagonal direction between the second direction D2 and the first direction D1.

[0016] Based on the above, the above embodiment provides an image processing device that can acquire a converted image in which fine lines are appropriately connected when the resolution conversion ratio differs depending on the orientation.

[0017] Here, the conversion of the resolution of the input image means a conversion that performs at least one of two actions: enlargement, which increases the number of pixels, and reduction, which decreases the number of pixels, in at least one of the first and second directions. Therefore, the conversion of the resolution of the input image includes, for example, a conversion that changes the number of pixels in the first direction but does not change the number of pixels in the second direction, a conversion that increases the number of pixels in the first direction but decreases the number of pixels in the second direction, and so on. In this application, "first," "second," etc., are terms used to identify each component included in a group of similar components, and do not imply any order. Which component among the group of components corresponds to "first," "second," etc., is determined relatively. For example, if multiple first pixels of an input image are arranged in the X and Y directions, when the X direction is fitted to the first direction, the Y direction is fitted to the second direction, and when the Y direction is fitted to the first direction, the X direction is fitted to the second direction. The first conversion ratio refers to the ratio of the number of pixels in the first direction of the converted image to the number of pixels in the first direction of the input image. If the first conversion ratio is less than 1, a resolution conversion is performed that reduces the number of pixels in the first direction. If the first conversion ratio is greater than 1, a resolution conversion is performed that increases the number of pixels in the first direction. The second conversion ratio refers to the ratio of the number of pixels in the second direction of the converted image to the number of pixels in the second direction of the input image. If the second conversion ratio is less than 1, a resolution conversion is performed that reduces the number of pixels in the second direction. If the second conversion ratio is greater than 1, a resolution conversion is performed that increases the number of pixels in the second direction. The first filter size refers to the number of filter coefficients in the first direction of the Gaussian filter. The second filter size refers to the number of filter coefficients in the second direction of the Gaussian filter. The condition that the standard deviation σ2 matches the standard deviation σ1 does not mean that the standard deviations σ1 and σ2 obtained from the filter coefficients are exactly the same, but also that they are approximately the same. For example, the filter coefficients obtained from standard deviations σ1 and σ2 where σ1=σ2 will have different errors depending on the filter size, etc. Therefore, it is sufficient that the standard deviations σ1 and σ2 obtained from the filter coefficients are approximately the same. Furthermore, the above-mentioned supplementary statement also applies in the following embodiments.

[0018] [Aspect 2] As illustrated in Figure 9, the blurring processing unit U1 may generate the blurred image IM11 when at least one of the first conversion magnification R1 and the second conversion magnification R2 is less than 1. The resolution conversion unit U2 may generate the converted image IM2 from the input image IM1 when the resolution is converted by the first conversion magnification R1 in the first direction D1 and the second conversion magnification R2 in the second direction D2, when the first conversion magnification R1 and the second conversion magnification R2 are 1 or greater. When a resolution conversion is performed that increases the number of pixels, the impact on the image quality of the converted image IM2 is minimal even without blurring. Therefore, the above configuration can shorten the processing time when enlarging the image.

[0019] [Aspect 3] As illustrated in Figure 11, the blurring processing unit U1 may set the first filter size S1 of the Gaussian filter F0 to an odd number greater than the reciprocal of the first conversion magnification R1 when the first conversion magnification R1 is less than 1. The blurring processing unit U1 may also set the second filter size S2 of the Gaussian filter F0 to an odd number greater than the reciprocal of the second conversion magnification R2 when the second conversion magnification R2 is less than 1. In this embodiment, since the information of pixels PX1 of the input image IM1 is not lost during image reduction, a high-quality converted image can be obtained.

[0020] [Aspect 4] Incidentally, a printing system SY1 according to one aspect of this technology, as illustrated in Figure 1, comprises the image processing device U0 described above, a print head 220 having a plurality of nozzles capable of ejecting liquid onto a medium ME0, and a control unit U10 that controls the ejection of the liquid from the print head 220 so that a printed image IM3 based on the converted image IM2 is formed on the medium ME0. This aspect provides a printing system capable of acquiring a converted image in which fine lines are appropriately connected when the resolution conversion ratio differs between the first and second directions.

[0021] [Aspect 5] As illustrated in Figures 1 and 10, the printing system SY1 may further include a resolution receiving unit U12 that accepts the setting of the resolution of the converted image IM2 from among the multiple output resolutions RE in a printing condition C0 in which the resolution of the converted image IM2 can be set to one of the multiple output resolutions RE. As illustrated in Figure 8, the output resolution RE may include a first output resolution RE1 and a second output resolution RE2 that is different from the first output resolution RE1. The blurring processing unit U1 may match the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the second output resolution RE2 is accepted by the resolution receiving unit U12 to the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the first output resolution RE1 is accepted by the resolution receiving unit U12.

[0022] In print condition C0, where the resolution of the converted image IM2 can be set to multiple different output resolutions RE, the standard deviations σ1 and σ2 of the Gaussian filter F0 remain unchanged regardless of whether the first output resolution RE1 or the second output resolution RE2 is used.

[0023] Under the above printing condition C0, it is assumed that if the resolution in the first direction D1 remains unchanged, the resolution in the second direction D2 will change. As illustrated in Figure 2, when the second conversion magnification R2 of the resolution in the second direction D2 is greater than the first conversion magnification R1 of the resolution in the first direction D1, and a Gaussian filter F0 with a first filter size S1 greater than the second filter size S2 is applied to the input image IM1, the following effect is obtained. Because the blur range in the first direction D1 is wider than the blur range in the second direction D2, the thickness of the thin line L2 along the second direction D2 is maintained, as illustrated in Figure 7. Also, because the blur range in the second direction D2 is narrower than the blur range in the first direction D1, excessive blurring does not occur in the second direction D2, where the conversion ratio is relatively large. Furthermore, since the standard deviation σ2 of the second direction D2 matches the standard deviation σ1 of the first direction D1 in the Gaussian filter F0, the blur effect in the first direction D1 and the second direction D2 is appropriate, as is the blur effect in the diagonal direction between the first direction D1 and the second direction D2.

[0024] As illustrated in Figure 3, when the second conversion magnification R2 of the resolution in the second direction D2 is smaller than the first conversion magnification R1 of the resolution in the first direction D1, and a Gaussian filter F0 with a first filter size S1 smaller than the second filter size S2 is applied to the input image IM1, the following effect is obtained. Because the blur range in the first direction D1 is narrower than the blur range in the second direction D2, the thin line L2 along the second direction D2 does not become too thick, and the first direction D1, where the conversion ratio is relatively large, does not become too blurred. As a result, the thin line L2 along the second direction D2 does not become too faint. In addition, in the Gaussian filter F0, the standard deviation σ2 of the second direction D2 matches the standard deviation σ1 of the first direction D1, so the blur effect in the first direction D1 and the second direction D2 is appropriate, and the blur effect in the diagonal direction between the first direction D1 and the second direction D2 is also appropriate.

[0025] Based on the above, the above embodiment makes it possible to achieve a suitable blurring effect and standardize the line widths in the printed image even when the output resolution changes under the set printing conditions.

[0026] [Aspect 6] As illustrated in Figure 10, the printing system SY1 may further include a printing condition receiving unit U11 that accepts the setting of the printing condition C0 from among a plurality of candidates. The blurring processing unit U1 may apply the Gaussian filter F0 to the input image IM1 using the standard deviations σ1 and σ2 from among the standard deviations σi associated with each of the plurality of candidates that are associated with the printing condition C0. This embodiment makes it possible to obtain a high-quality printed image according to the printing condition set from among a plurality of candidates.

[0027] [Aspect 7] As illustrated in Figure 8, the medium ME0 may include a first medium ME1 and a second medium ME2 that is less permeable to the liquid than the first medium ME1. The plurality of candidates that can be the printing condition C0 may include a first medium candidate CM1 that forms the printed image IM3 on the first medium ME1, and a second medium candidate CM2 that forms the printed image IM3 on the second medium ME2. The blurring processing unit U1 may make the standard deviations σ1,σ2 of the Gaussian filter F0 applied to the input image IM1 when the second medium candidate CM2 is accepted as the printing condition C0 larger than the standard deviations σ1,σ2 of the Gaussian filter F0 applied to the input image IM1 when the first medium candidate CM1 is accepted as the printing condition C0. This embodiment can obtain a high-quality printed image according to the type of medium.

[0028] [Aspect 8] As illustrated in Figure 8, the plurality of candidates that can be the printing condition C0 may include a first image candidate C11 that prioritizes the sharpness of the line drawings contained in the input image IM1, and a second image candidate C12 that prioritizes the gradation of the input image IM1. The blurring processing unit U1 may make the standard deviations σ1,σ2 of the Gaussian filter F0 applied to the input image IM1 when the second image candidate C12 is accepted as the printing condition C0 larger than the standard deviations σ1,σ2 of the Gaussian filter F0 applied to the input image IM1 when the first image candidate C11 is accepted as the printing condition C0. This embodiment can obtain a high-quality printed image according to the type of input image.

[0029] [Aspect 9] As illustrated in Figure 10, the blurring processing unit U1 may accept an operation on the user interface screen 500 to set the standard deviation σ of the Gaussian filter F0 applied to the input image IM1. The blurring processing unit U1 may then apply the Gaussian filter F0 having the filter coefficients KE, with the accepted standard deviations σ as the standard deviations σ1 and σ2, to the input image IM1. This embodiment allows for obtaining a print image that is more to the user's liking.

[0030] [Aspect 10] Furthermore, an image processing method according to one aspect of this technology is an image processing method that obtains a converted image IM2 whose resolution has been converted based on an input image IM1 having a plurality of pixels PX1 arranged in a first direction D1 and a second direction D2 intersecting the first direction D1, and includes the following steps (A) and (B). (A) Blurring process ST1, which generates a blurred image IM11 by applying a two-dimensional Gaussian filter F0 having a first filter size S1 in the first direction D1 and a second filter size S2 in the second direction D2 to the input image IM1. (B) A resolution conversion step ST2 in which the resolution is converted by a first conversion ratio R1, which is the resolution conversion ratio in the first direction D1, and a second conversion ratio R2, which is the resolution conversion ratio in the second direction D2, and the converted image IM2 is generated from the blurred image IM11. Here, in the blurring process ST1, if the first filter size S1 is made larger than the second filter size S2, the second conversion magnification R2 is larger than the first conversion magnification R1. In the blurring process ST1, if the second filter size S2 is made larger than the first filter size S1, the first conversion magnification R1 is larger than the second conversion magnification R2. Furthermore, the standard deviation σ of the Gaussian filter F0 in the first direction D1 is defined as standard deviation σ1, and the standard deviation σ of the Gaussian filter F0 in the second direction D2 is defined as standard deviation σ2. In this image processing method, in the blurring process ST1, if the second filter size S2 is different from the first filter size S1, the Gaussian filter F0 having a filter coefficient KE in which the standard deviation σ2 matches the standard deviation σ1 is applied to the input image IM1. The above embodiment provides an image processing method that can obtain a converted image in which thin lines are properly connected when the resolution conversion ratio differs depending on the orientation.

[0031] Furthermore, this technology is applicable to a composite device including the image processing device described above, a printing method for the printing system described above, an image processing program for implementing the image processing method described above on a computer, a printing control program for the printing system described above, a computer-readable recording medium on which any of the aforementioned control programs are recorded, and so on. Any of the aforementioned devices may consist of multiple distributed parts.

[0032] (2) Specific examples of the configuration of a printing system including an image processing device: Figure 1 schematically shows an example of a printing system configuration including an image processing device. Figures 2 and 3 schematically show an example of generating a converted image after performing a blurring process using a Gaussian filter. The printing system SY1 shown in Figure 1 includes a host device 100 and a printer 200, and is capable of forming a printed image IM3 on a medium ME0 based on a converted image IM2. The host device 100 includes a processor CPU 111, ROM 112, RAM 113, storage device 114, input device 115, display device 116, communication I / F 117, etc. Here, CPU is an abbreviation for Central Processing Unit, ROM is an abbreviation for Read Only Memory, RAM is an abbreviation for Random Access Memory, and I / F is an abbreviation for Interface. The aforementioned elements (111 to 117) are electrically connected and capable of inputting and outputting information to each other. The ROM 112, RAM 113, and storage device 114 are memories, and at least the ROM 112 and RAM 113 are semiconductor memories.

[0033] The storage device 114 stores an OS (not shown), an image processing program PR0, a Gaussian filter F0 used for blurring shown in Figures 2 and 3, a print control program PR1, etc. Here, OS is an abbreviation for operating system. The storage device 114 can be a non-volatile semiconductor memory such as flash memory, a magnetic storage device such as a hard disk, etc. The input device 115 can be a pointing device, a hard key including a keyboard, a touch panel attached to the surface of the display panel, etc. The display device 116 displays a screen corresponding to the display information based on the display information. The display device 116 can be a liquid crystal display panel, etc. The communication I / F 117 is connected to the communication I / F 230 of the printer 200 and inputs and outputs information such as print data to the printer 200. Communication between the communication I / F 117 and 230 may be wired, wireless, or a network communication such as a LAN or the Internet. Here, LAN is an abbreviation for Local Area Network.

[0034] The image processing program PR0 shown in Figure 1 implements a blurring function FU1 and a resolution conversion function FU2 in the host device 100. The print control program PR1 shown in Figure 1 implements a print condition acceptance function FU11, a resolution acceptance function FU12, a color conversion function FU3, and a halftone processing function FU4 in the host device 100. The CPU 111 of the host device 100 reads information stored in the storage device 114 into the RAM 113 as appropriate and performs various processing by executing the read program. The CPU 111 performs processing corresponding to the above-mentioned functions (FU1~FU4, FU11, FU12) by executing the programs (PR0, PR1) read into the RAM 113. The image processing program PR0 causes the host device 100, which is a computer, to function as an image processing device U0 equipped with a blurring processing unit U1 and a resolution conversion unit U2. The print control program PR1 causes the host device 100 to function as a print condition receiving unit U11, a resolution receiving unit U12, a color conversion unit U3, and a halftone processing unit U4. Furthermore, the host device 100, which executes the image processing program PR0, performs a blurring process ST1 and a resolution conversion process ST2, as illustrated in Figures 9 and 11. The host device 100, which executes the print control program PR1, performs the print condition acceptance process ST11, the resolution acceptance process ST12, the color conversion process ST3, and the halftone processing process ST4, as illustrated in Figure 9. The computer-readable recording medium storing the programs (PR0, PR1) that enable the computer to implement the above-mentioned functions (FU1~FU4, FU11, FU12) is not limited to the internal storage device of the host device, but may also be an external recording medium.

[0035] The host device 100 includes computers such as personal computers, mobile phones such as smartphones, digital cameras, digital video cameras, etc. The host device 100 may have all its components (111-117) in a single enclosure, or it may be composed of multiple devices that are separated and able to communicate with each other. Furthermore, this technology can be implemented even if at least a part of the printer 200 is located in the host device 100.

[0036] The printer 200 shown in Figure 1 is an inkjet printer that forms a printed image IM3 corresponding to the print data by ejecting C ink, M ink, Y ink, and K ink as colorants from the print head 220. Here, ink is an example of a liquid, C means cyan, M means magenta, Y means yellow, and K means black. Of course, the printer 200 may also be an electrophotographic printer such as a laser printer that uses toner, a 3D printer, etc. The print head 220 has a plurality of nozzles Nc capable of ejecting C ink droplets onto the medium ME0, a plurality of nozzles Nm capable of ejecting M ink droplets onto the medium ME0, a plurality of nozzles Ny capable of ejecting Y ink droplets onto the medium ME0, and a plurality of nozzles Nk capable of ejecting K ink droplets onto the medium ME0. The print head 220 is supplied with C, M, Y, and K inks from ink cartridges Cc, Cm, Cy, and Ck, respectively. As a result, the print head 220 ejects ink droplets 280 of C, M, Y, and K from nozzles Nc, Nm, Ny, and Nk, respectively. When the ink droplets 280 land on the medium ME0, ink dots are formed on the medium ME0. As a result, a printed material with a printed image IM3 on the medium ME0 is obtained. The medium ME0 is not particularly limited and includes paper, fabric, resin, metal, etc. The shape of the medium ME0 may be in the form of a roll, a cut two-dimensional shape, or a three-dimensional shape.

[0037] The printer 200 includes a controller 210 that controls the ejection of ink from the print head 220. The controller 210, together with the host device 100 that executes the print control program PR1, constitutes a control unit U10 that controls the ejection of ink from the print head 220 so that a print image IM3 based on the converted image IM2 is formed on the medium ME0.

[0038] As shown in Figures 2 and 3, the input image IM1 has multiple pixels PX1 arranged in the X and Y directions. The X and Y directions are assumed to be orthogonal, but they do not need to be orthogonal as long as they intersect. In the example shown in Figures 2 and 3, the X direction is mapped to the first direction D1, and the Y direction is mapped to the second direction D2. It is also possible to map the Y direction to the first direction D1 and the X direction to the second direction D2. Figures 2 and 3 show that in the schematic example of the input image IM1, X1 pixels PX1 are arranged in the X direction, and Y1 pixels PX1 are arranged in the Y direction. If the color system of the input image IM1 is RGB, the pixel value of each pixel PX1 is a combination of R, G, and B values. Here, R means red, G means green, and B means blue. The R, G, and B values ​​are represented, for example, by grayscale values ​​from 0 to 255.

[0039] The blurring unit U1 included in the image processing device U0 can generate a blurred image IM11 by applying a two-dimensional Gaussian filter F0 of size S1×S2 to the input image IM1. The S1×S2 Gaussian filter F0 means that it has a first filter size S1 in the X direction and a second filter size S2 in the Y direction. The first filter size S1 means the number of filter coefficients KE in the X direction, and the second filter size S2 means the number of filter coefficients KE in the Y direction. The filter coefficients KE are also called kernels. The fact that the Gaussian filter F0 is two-dimensional means that both the first filter size S1 and the second filter size S2 are multiple. The blurring unit U1 sequentially sets a target pixel PX1t from among multiple pixels PX1, and generates a blurred image IM11 by applying the Gaussian filter F0 to the input image IM1 centered on the target pixel PX1t. Here, let σ1 be the standard deviation σ of the Gaussian filter F0 in the X direction, and σ2 be the standard deviation σ of the Gaussian filter F0 in the Y direction. The Gaussian filter F0 shown in Figures 2 and 3 has the characteristic that σ1 = σ2, even though the filter sizes (S1, S2) are different in the X and Y directions.

[0040] The resolution conversion unit U2 included in the image processing device U0 can generate a converted image IM2 from a blurred image IM11, with the resolution converted by a conversion ratio of R1 × R2. The conversion ratio of R1 × R2 means that the first conversion ratio R1 is the resolution conversion ratio in the X direction, and the second conversion ratio R2 is the resolution conversion ratio in the Y direction. Here, assume that the resolution of the input image IM1 is RX1 × RY1 dpi, and the resolution of the converted image IM2 is RX2 × RY2 dpi. The resolution of the blurred image IM11 is RX1 × RY1 dpi. The first conversion ratio R1 is the ratio RX2 / RX1 of the resolution RX2 of the converted image IM2 in the X direction to the resolution RX1 of the input image IM1 and the blurred image IM11 in the X direction, and is greater than 0. The second conversion magnification R2 is the ratio RY2 / RY1 of the resolution RY1 of the input image IM1 and the blurred image IM11 in the Y direction to the resolution RY2 of the converted image IM2 in the Y direction, and is greater than 0. In the schematic example shown in Figures 2 and 3, each pixel in the converted image IM2 will be called a second pixel PX2. Assume that in the converted image IM2, X2 second pixels PX2 are arranged in the X direction and Y2 second pixels PX2 are arranged in the Y direction. The first conversion magnification R1 in the X direction is also X2 / X1, and the second conversion magnification R2 in the Y direction is also Y2 / Y1.

[0041] Generally, the standard deviation σ of a Gaussian filter used for blurring before resolution conversion is increased as the conversion magnification decreases. The recommended range for the standard deviation σ to reduce aliasing (moire) is from 1 / (conversion magnification × π) to 3 / (conversion magnification × π). An example is shown below. For a conversion ratio of 0.75, the recommended standard deviation σ is 0.42 to 1.27. For a conversion ratio of 0.50, the recommended standard deviation σ is 0.64 to 1.91. For a conversion ratio of 0.25, the recommended standard deviation σ is 1.27 to 3.82. The Gaussian filter F0 in this specific example prioritizes maintaining the continuity and thickness of fine lines in the converted image IM2, and, assuming that the filter size (S1, S2) is set separately for the X and Y directions, it has the characteristic that the standard deviation is σ1 = σ2 regardless of the conversion magnification (R1, R2). Furthermore, the Gaussian filter F0 in this specific example has the characteristic that, under printing conditions C0 where the resolution of the converted image IM2 can be set to multiple different output resolutions RE, the standard deviation remains σ1 = σ2 even when the conversion magnification (R1, R2) changes due to the change in output resolution RE.

[0042] The blurring processing unit U1 may apply a Gaussian filter F0 to the input image IM1 only if R1 < 1 or R1 < 2. R1 < 1 means the image is reduced in the X direction, and R2 < 1 means the image is reduced in the Y direction. If blurring is not performed before the image is reduced in at least one of the X and Y directions, pixels PX1 with less information will be reflected in the resolution conversion. Even when interpolation is performed that references multiple pixels PX1 during the resolution conversion, the weight of the pixel value of a pixel PX1 that is far from the reference point corresponding to the second pixel PX2 included in the converted image IM2 is small. Therefore, the blurring processing unit U1 performs blurring when the image is reduced in at least one of the X and Y directions. On the other hand, if the image is not reduced in both the X and Y directions, the information of all pixels PX1 will be reflected in the resolution conversion even if blurring is not performed before the resolution conversion. Therefore, the blurring processing unit U1 does not need to perform blurring when R1 ≥ 1 and R2 ≥ 1. The resolution conversion unit U2 may generate a converted image IM2 from the input image IM1 when R1≧1 and R2≧1.

[0043] The resolution conversion unit U2 sequentially selects a target pixel PX2t from among a plurality of second pixels PX2 that will become the converted image IM2, and determines the pixel value of the target pixel PX2t by referencing multiple pixels PX1 with reference points corresponding to the target pixel PX2t. For example, the resolution conversion unit U2 determines the pixel value of the second pixel PX2 by performing an interpolation operation using the pixel values ​​of multiple pixels PX1 within a predetermined range with reference points in the blurred image IM11 or input image IM1. The reference points are set on the XY coordinate plane aligned with the blurred image IM11 or input image IM1 in order to determine the pixel value of each second pixel PX2. For the interpolation operation, a bicubic interpolation operation can be used, which references up to 4 × 4 pixels PX1 with reference points. For reference points that are outside the blurred image IM11, etc., among the 4 × 4 reference points with reference points, it is assumed that pixels with the pixel values ​​of pixels PX1 at the edges of the blurred image IM11, etc., exist outside the blurred image IM11, etc., and those pixel values ​​are applied. Of course, the interpolation operation is not limited to the bicubic method; it may also be a bilinear method that references up to 2x2 pixels PX1 based on a reference point.

[0044] The resolution of the printed image IM3 formed by printer 200 may differ in the X and Y directions. Therefore, the resolution of the converted image IM2 used to form the printed image IM3 may also differ in the X and Y directions. If the resolution conversion ratios (R1, R2) differ between the X and Y directions, and blurring is not performed properly, the fine lines contained in the input image IM1 may be interrupted after the resolution conversion. If the resolution changes in steps, and blurring is not performed properly, the thickness of the fine lines contained in the input image IM1 may change according to the resolution. Therefore, the blurring processing unit U1 does not limit the filter size of the Gaussian filter F0 to S1=S2, and even if the filter sizes (S1,S2) are different in the X and Y directions, it sets the standard deviation of the Gaussian filter F0 to σ1=σ2.

[0045] Figure 2 schematically illustrates how a blurring process using a horizontal Gaussian filter F0 is performed to generate a converted image IM2 in which the vertical conversion ratio (R2) is greater than the horizontal conversion ratio (R1). Figure 3 schematically illustrates how a blurring process using a vertical Gaussian filter F0 is performed to generate a converted image IM2 in which the horizontal conversion ratio (R1) is greater than the vertical conversion ratio (R2). As shown in Figure 2, when the blur processing unit U1 sets the filter size to S1 > S2, the second conversion magnification R2 is greater than the first conversion magnification R1. This is expressed in this way because, when R2 > R1, S1 = S2 can occur depending on the conversion magnification. Of course, when R2 > R1, S1<S2とはならず、S1> S2 is possible. As shown in Figure 3, when the blur processing unit U1 sets the filter size to S2 > S1, the first conversion magnification R1 is greater than the second conversion magnification R2. This is expressed in this way because, when R1 > R2, S2 = S1 can occur depending on the conversion magnification. Of course, when R1 > R2, S2<S1とはならず、S2> It can be S1. In either case, the blurring processing unit U1 applies a Gaussian filter F0 to the input image IM1, which has filter coefficients KE whose standard deviation σ2 matches the standard deviation σ1, when the second filter size S2 is different from the first filter size S1.

[0046] Figure 4 schematically shows an example of the relationship between the resolution conversion ratio (R1, R2) and the filter size (S1, S2) of the Gaussian filter F0. The Gaussian filters F11 to F15 shown in Figure 4 are specific examples of the Gaussian filter F0. In Gaussian filter F13, the filter coefficients a, b, and c are shown as the respective filter coefficients KE, while in Gaussian filters F11 and F15, the filter coefficients d, e, f, g, h, i, and j are shown as the respective filter coefficients KE. In a Gaussian filter F0, each filter coefficient KE is assumed to be equally spaced in both the X direction (D1) and the Y direction (D2). The coordinates (x,y) of each filter coefficient KE change by 1 in both the X direction and the Y direction, with the center of the Gaussian filter F0 being (0,0). For example, a 7x3 Gaussian filter F15 has filter coefficients KE in the range of coordinates (-3,-1) to (3,1), and a 3x7 Gaussian filter F11 has filter coefficients KE in the range of coordinates (-1,-3) to (1,3).

[0047] In a Gaussian filter F0, when both the standard deviation σ1 in the X direction and the standard deviation σ2 in the Y direction are standard deviation σ, the Gaussian distribution function representing the weight of each filter coefficient KE is expressed by the following equation.

number

number

[0048] According to the Gaussian distribution function, for the filter coefficients a, b, c of the 3×3 Gaussian filter F13, 0 < c < b < a. For the filter coefficients d, e, f, g, h, i, j of the 7×3 Gaussian filter F15 and the 3×7 Gaussian filter F11, 0 < j < i < h < g < f < e < d. When the standard deviation σ is the same for the Gaussian filter F11 and the Gaussian filter F15, the ratio of the filter coefficients f, e, d in the part surrounded by the thick line in the Gaussian filters F11, F15 approximately matches the ratio of the filter coefficients c, b, a of the Gaussian filter F13. f:e:d ≒ c:b:a The expression "approximately matches" is because even if the standard deviation σ for determining the filter coefficient KE is the same for the Gaussian filters F11 to F15, errors occur due to calculations such as normalization and integerization from the Gaussian distribution function to determine the filter coefficient KE. Therefore, the fact that the standard deviation σ of a certain Gaussian filter matches the standard deviation σ of another Gaussian filter with a different filter size is not limited to the case where the standard deviation σ obtained from the filter coefficient KE is exactly the same, but includes the case where the standard deviation σ obtained from the filter coefficient KE is approximately the same. For example, when the filter coefficients a, b, c of the Gaussian filter F13 and the filter coefficients d, e, f, g, h, i, j of the Gaussian filters F11, F15 are obtained from a certain standard deviation σ, it can be said that the standard deviation σ of these Gaussian filters is the same.

[0049] Furthermore, Gaussian filters F11 to F15 all have the same standard deviation σ1 in the X direction and σ2 in the Y direction. σ1=σ2 means that the filter coefficients KE arranged in the X direction passing through the center of the Gaussian filter F0, and the filter coefficients KE arranged in the Y direction passing through the center of the Gaussian filter F0, can be determined from the standard deviations σ=σ1=σ2. Therefore, the fact that the standard deviation σ2 matches the standard deviation σ1 does not mean that the standard deviations σ1 and σ2 obtained from the filter coefficients KE are exactly the same, but also that the standard deviations σ1 and σ2 obtained from the filter coefficients KE are approximately the same. This is because errors occur when determining the filter coefficients KE by normalizing or integerizing the standard deviation σ from the Gaussian distribution function using σ1=σ2. For example, in a 7x3 Gaussian filter F15, if the filter coefficients i, g, e, d, e, g, i aligned in the X direction and the filter coefficients e, d, e aligned in the Y direction can be determined from a certain standard deviation σ = σ1 = σ2, then it can be said that the standard deviation σ2 is equal to the standard deviation σ1.

[0050] The filter sizes (S1, S2) of the Gaussian filter F0 shown in Figure 4 are set as follows. If the first transformation magnification R1 in the X direction is less than 1, the first filter size S1 in the X direction is set to the smallest odd number greater than the reciprocal of the first transformation magnification, 1 / R1. If the second transformation magnification R2 in the Y direction is less than 1, the second filter size S2 in the Y direction is set to the smallest odd number greater than the reciprocal of the second transformation magnification, 1 / R2.

[0051] For example, if R1 = 0.5, then 1 / R1 = 2, so S1 = 3. If R1 = 0.3, then 1 / R1 = 3.3, so S1 = 5. If R1 = 0.15, then 1 / R1 = 6.7, so S1 = 7. Similarly, if R2 = 0.5, then S2 = 3; if R2 = 0.3, then S2 = 5; and if R2 = 0.15, then S2 = 7.

[0052] Figure 12 schematically illustrates the filter coefficients KE of Gaussian filters F11 to F14 when σ=1, for reference. In each Gaussian filter F0, the sum of all filter coefficients KE is 100%, and each filter coefficient KE is shown as a percentage to one decimal place. Although not shown in the figure, each filter coefficient KE of the 7×3 Gaussian filter F15 is the same as each filter coefficient KE of the 3×7 Gaussian filter F11 rotated by 90°. For example, the filter coefficients f:e:d = 6.6%:10.9%:18.0% of a 3x7 Gaussian filter F11 are roughly equivalent to the filter coefficients c:b:a = 7.5%:12.4%:20.4% of a 3x3 Gaussian filter F13.

[0053] Figure 5 schematically illustrates the appearance of each image when a blurring process is performed using a horizontal Gaussian filter F0, and then a converted image IM2 is generated in which the vertical conversion ratio (R2) is greater than the horizontal conversion ratio (R1). The input image IM1 contains thin lines L1 along the X direction (D1) and thin lines L2 along the Y direction (D2). These lines L1 and L2 are darker than their surroundings, like black, and intersect in a cross shape. When the filter size of the Gaussian filter F0 is S1 > S2, the blur range in the X direction is wider than the blur range in the Y direction. In the blurred image IM11 obtained by applying a horizontally elongated Gaussian filter F0 to the input image IM1, the line L2 along the Y direction is thinner and wider than the line L1 along the X direction. When this blurred image IM11 is converted with a conversion ratio such that the resolution R2 > R1, the thickness of the line L1 along the X direction and the thickness of the line L2 along the Y direction become the same.

[0054] Although not shown, the same can be said when generating a transformed image IM2 in which the horizontal transformation magnification (R1) is greater than the vertical transformation magnification (R2) after performing blurring processing using a vertically long Gaussian filter F0. When the filter size of the Gaussian filter F0 is S2 > S1, the blurring range in the Y direction is wider than the blurring range in the X direction. In the blurred image IM11 obtained by applying the vertically long Gaussian filter F0 to the input image IM1, the line L1 along the X direction becomes a thinner and more spread-out line compared to the line L2 along the Y direction. When this blurred image IM11 is transformed at a transformation magnification where R1 > R2, the thickness of the line L1 along the X direction and the thickness of the line L2 along the Y direction become equal.

[0055] FIG. 6 schematically illustrates a transformed image IM2 whose resolution has been transformed after performing blurring processing on an input image IM1 having thin lines in a direction closer to the Y direction (D2) than the X direction (D1) using a vertically long Gaussian filter F0. The thin lines included in the input image IM1 are darker than the surrounding area, like black. The transformation magnification of the resolution is such that the vertical transformation magnification (R2) is smaller than the horizontal transformation magnification (R1). On the left side of FIG. 6, a transformed image IM92 obtained by applying a 3×3 Gaussian filter to the aforementioned input image IM1 and then performing a resolution transformation of 0 < R1 < 1 / 3 (for example, R1 = 0.25) is illustrated. The diagonal line L90 included in the transformed image IM92 is broken in a dashed line. This is because, during the resolution transformation, almost all the information of some pixels PX1 included in the input image IM1 in the Y direction is lost. Incidentally, if a Gaussian filter with a large filter size such as 7×7 is applied to the input image IM1 instead of the 3×3 Gaussian filter, the blurring range becomes too wide in the X direction and the image is overly blurred, resulting in a degradation of the image quality of the transformed image IM2.

[0056] On the other hand, the diagonal line L0 included in the transformed image IM2 whose resolution has been transformed after performing blurring processing using a 3×7 Gaussian filter F0 is connected. This is because the loss of information of the pixels PX1 included in the input image IM1 in the Y direction is suppressed during the resolution transformation. Although not shown, the same can be said when blurring processing is performed on the input image IM1 having thin lines in a direction closer to the X direction (D1) than the Y direction (D2) using the horizontally long Gaussian filter F0, and then resolution conversion of R1 < R2 is performed. The diagonal line L0 included in the obtained converted image IM2 has its connection ensured. This is because information loss of the pixel PX1 included in the input image IM1 in the X direction is suppressed during resolution conversion.

[0057] FIG. 7 schematically illustrates a converted image IM2 obtained after performing blurring processing by applying the Gaussian filter F0 of S1×S2 with a standard deviation σ1 = σ2 that does not change even when the resolution in the Y direction is changed to the input image IM1. FIG. 13 schematically illustrates, as a comparative example, a converted image IM2 obtained after performing blurring processing by applying a square Gaussian filter whose standard deviation σ increases as the resolution decreases according to a general recommended value to the input image IM1. Note that the input image IM1 includes thin lines along the Y direction (D2). The vertical thin lines included in the input image IM1 are darker than the surroundings, like black. In the comparative example shown in FIG. 13, when the resolution in the Y direction becomes equal to or less than the resolution in the X direction, as the resolution in the Y direction decreases, the vertical line L92 becomes thinner and thicker. On the other hand, as shown in FIG. 7, when the Gaussian filter F0 of S1×S2 with a standard deviation σ1 = σ2 that does not change even when the resolution in the Y direction is changed is used, the thickness of the line L2 hardly changes even when the resolution in the Y direction changes, and the darkness of the line L2 also hardly changes. Although not shown, when the Gaussian filter F0 of S1×S2 with a standard deviation σ1 = σ2 that does not change even when the resolution in the X direction is changed is used, the thickness of the line along the X direction hardly changes, and the darkness of the line also hardly changes.

[0058] Note that, as a smoothing filter, a moving average filter with all filter coefficients being the same is known. The fact that all filter coefficients are the same means that not only the filter coefficients of the coordinates shifted by 1 in the X direction and the Y direction from the center of the moving average filter, but also, for example, diagonally by 2 from the center of the moving average filter 1 / 2This means that the filter coefficients for coordinates shifted by approximately 1.4 will also have the same value. Therefore, when a moving average filter is applied to the input image IM1, the diagonal components are overemphasized, resulting in a decrease in the image quality of the transformed image IM2. The Gaussian filter F0 described above has a standard deviation σ2 in the Y direction (D2) that matches the standard deviation σ1 in the X direction (D1), so the blurring effect in the X and Y directions is moderate, as is the blurring effect in the diagonal direction between the X and Y directions.

[0059] The standard deviations σ1 and σ2 of the Gaussian filter F0 are set according to the printing conditions C0, which allow the resolution of the converted image IM2 to be set to one of several different output resolutions RE, as illustrated in Figure 8. Figure 8 schematically illustrates the standard deviations σi associated with each printing condition C0. The standard deviations σi shown in Figure 8 are merely examples, and the standard deviations σi associated with each printing condition C0 can be changed as appropriate. When print condition C0 is set, a Gaussian filter F0 is applied to the input image IM1 using the standard deviation σi associated with print condition C0, or the standard deviation adjusted from σi as standard deviations σ1 and σ2.

[0060] The standard deviation σi shown in Figure 8 is set according to the type of medium ME0 or input image IM1. Figure 8 can also be said to show multiple candidates that may be the printing condition C0.

[0061] From the perspective of the medium ME0, the multiple candidates shown in Figure 8 include cloth, plain paper, and photographic paper. Cloth is associated with a standard deviation σi of 0.5, plain paper with a standard deviation σi of 1.5, and photographic paper with a standard deviation σi of 2.0. Assuming that the standard deviation σi=2.0 for photographic paper is the default, plain paper is more prone to bleeding than photographic paper and requires a slightly sharper image, so its standard deviation σi is set to a slightly smaller 1.5. Cloth is more prone to bleeding than plain paper and requires an even sharper image, so its standard deviation σi is set to an even smaller 0.5. For example, if plain paper is assigned to the first medium ME1, then photographic paper is assigned to the second medium ME2 because photographic paper is less prone to ink bleeding than plain paper. In this case, plain paper is assigned to the first medium candidate CM1 for forming the printed image IM3 on the first medium ME1, and photographic paper is assigned to the second medium candidate CM2 for forming the printed image IM3 on the second medium ME2. Furthermore, if cloth is assigned to the first medium ME1, then plain paper or photographic paper is assigned to the second medium ME2.

[0062] From the perspective of the input image IM1, the multiple candidates shown in Figure 8 include "line drawings" and "natural images and graphics." A standard deviation σi of 1.0 is associated with "line drawings," and a standard deviation σi of 2.0 is associated with "natural images and graphics." The standard deviation σi is set to a small 1.0 for "line drawings" to prioritize the sharpness of the line drawings over that of "natural images and graphics." The standard deviation σi is set to a large 2.0 for "natural images and graphics" to balance the smoothness of the gradation with the sharpness. Based on the above, the first image candidate C11, which prioritizes the sharpness of the line drawings in the input image IM1, is "line drawing," and the second image candidate C12, which prioritizes the gradation of the input image IM1, is "natural image and graphic."

[0063] For any of the printing conditions C0 shown in Figure 8, the output resolution RE can be changed to 600×300dpi, 600×600dpi, 600×900dpi, ... Even if the output resolution RE is changed, the standard deviation σ1=σ2=σi applied to the Gaussian filter F0 remains unchanged. For example, if "Line drawing" is set as the printing condition C0 and a standard deviation σi=1.0 is applied to the Gaussian filter F0, the applied standard deviation σi=1.0 remains unchanged even if the output resolution RE is changed to 600×300dpi, 600×600dpi, 600×900dpi, ... Here, multiple output resolutions that are different from each other among 600×300dpi, 600×600dpi, 600×900dpi, ... are referred to as the first output resolution RE1 and the second output resolution RE2. As shown in Figure 8, when 600 × 300 dpi is assigned as the first output resolution RE1, then one of the following will be assigned as the second output resolution RE2: 600 × 600 dpi, 600 × 900 dpi, etc. Figure 8 shows that 600 × 900 dpi is assigned as the second output resolution RE2.

[0064] (3) Specific examples of print control processing: Figure 9 schematically illustrates the print control process in which a print image IM3 corresponding to a converted image IM2, whose resolution has been converted based on the input image IM1, is formed on the printer 200. Figure 10 schematically illustrates the UI screen 500 displayed in step S102 of the print control process. Here, UI is an abbreviation for user interface. Figure 11 schematically illustrates the blurred image generation process performed in step S110 of the print control process. The print control process will be explained below with reference to Figures 1 to 8. The print control process in this specific example is performed by the host device 100 shown in Figure 1. The print control process starts when the host device 100 receives a user operation at the input device 115 to form a print image IM3 on the printer 200. Here, steps S102 to S104 correspond to the print condition reception process ST11, the resolution reception process ST12, the print condition reception unit U11, the resolution reception unit U12, the print condition reception function FU12, and the resolution reception function FU12. Steps S104, S108 to S110 correspond to the blurring process ST1, the blurring processing unit U1, and the blurring function FU1. Steps S106 and S112 correspond to the resolution conversion process ST2, the resolution conversion unit U2, and the resolution conversion function FU2. Step S114 corresponds to the color conversion process ST3, the color conversion unit U3, and the color conversion function FU3. Step S116 corresponds to the halftone processing step ST4, the halftone processing unit U4, and the halftone processing function FU4. Hereafter, the term "step" may be omitted, and the step number may be indicated in parentheses.

[0065] When the print control process starts, the host device 100 displays the UI screen 500 shown in Figure 10 on the display device 116 (S102). The UI screen 500 includes a media type selection field 501, an image type selection field 502, a recommended standard deviation display field 503, a standard deviation adjustment area 504, a standard deviation input field 505, a resolution selection field 506, an OK button 507, and the like.

[0066] In the media type selection field 501, labeled "Media Type," the host device 100 accepts the setting of one type from among media candidates such as cloth, plain paper, and photographic paper. The user can select one type from among multiple media candidates by operating the media type selection field 501 with the input device 115. In the image type selection field 502, labeled "Image Type," the host device 100 accepts the setting of one type from among image candidates such as line drawings and nature images. The user can select one type from among multiple image candidates by operating the image type selection field 502 with the input device 115. The media type selection field 501 and the image type selection field 502 correspond to the print condition reception unit U11, which accepts the setting of print conditions C0 from among multiple candidates. The combination of the selected media candidate and the selected image candidate conforms to the set print conditions C0. The host device 100 displays, for example, the smaller of the standard deviation σi associated with the selected media candidate and the standard deviation σi associated with the selected image candidate in the recommended standard deviation display field 503 and the standard deviation input field 505. In the UI screen 500, the standard deviation σ is indicated as "blur amount". In the example shown in Figure 8, a standard deviation σi=1.5 is associated with plain paper and a standard deviation σi=1.0 is associated with line drawing, so when plain paper and line drawing are selected, a standard deviation σi=1.0 is displayed in the recommended standard deviation display field 503 and the standard deviation input field 505.

[0067] The standard deviation adjustment area 504 shown in Figure 10 has a slider control consisting of a horizontal slider bar and a slider. In the standard deviation adjustment area 504, the host device 100 increases or decreases the standard deviation σ applied to the Gaussian filter F0 from the recommended standard deviation σi by accepting the operation of the slider by the input device 115. The host device 100 displays the standard deviation σ according to the position of the slider in the standard deviation input field 505. In this standard deviation input field 505, the host device 100 accepts the input of the standard deviation σ applied to the Gaussian filter F0 by the input device 115. As described above, the host device 100 accepts an operation on the UI screen 500 to set the standard deviation σ of the Gaussian filter F0 applied to the input image IM1.

[0068] In the resolution selection field 506, labeled "Print Resolution," the host device 100 accepts the setting of the output resolution from among output resolutions RE such as 600×300dpi, 600×600dpi, and 600×900dpi. The user can select one of the multiple output resolutions RE by operating the resolution selection field 506 with the input device 115. The resolution selection field 506 corresponds to the resolution reception unit U12, which accepts the setting of the resolution of the converted image IM2 from among multiple output resolutions RE in print conditions C0, where the resolution of the converted image IM2 can be set to one of several different output resolutions RE.

[0069] When the host device 100 receives the OK button 507 via the input device 115, it acquires the set print conditions C0, the set output resolution, and the set standard deviation σ, and stores them in at least one of the RAM 113 and the storage device 114 (S104 in Figure 9). The print conditions C0 are a combination of the medium ME0 set in the medium type selection field 501 and the image set in the image type selection field 502.

[0070] Next, the host device 100 calculates the first conversion magnification R1 = RX2 / RX1 in the X direction and the second conversion magnification R2 = RY2 / RY1 in the Y direction based on the resolutions RX1, RY1, RX2, and RY2 before and after the resolution conversion (see Figures 2 and 3) (S106). After calculating the conversion ratios (R1, R2), the host device 100 determines whether R1 < 1 or R2 < 1 (S108). If R1 < 1 or R2 < 1, the host device 100 performs the blurred image generation process shown in Figure 11 (S110), and then proceeds to S112. Therefore, the blurring processing unit U1 generates a blurred image IM11 when at least one of the conversion ratios (R1, R2) is less than 1. If R1 ≥ 1 and R2 ≥ 1, the host device 100 proceeds to S112 without performing the blurred image generation process. Therefore, if both conversion ratios (R1, R2) are 1 or greater, the resolution conversion unit U2 generates a converted image IM2 from the input image IM1 with the resolution converted by the conversion ratios (R1, R2).

[0071] When the blurring image generation process shown in Figure 11 begins, the host device 100 calculates the first filter size S1 in the X direction and the second filter size S2 in the Y direction (see Figures 2 and 3) (S202). If R1 < 1, the host device 100 sets the first filter size S1 to the smallest odd number greater than the reciprocal of the first conversion magnification, 1 / R1. If R2 < 1, the host device 100 sets the second filter size S2 to the smallest odd number greater than the reciprocal of the second conversion magnification, 1 / R2. The filter sizes (S1, S2) are set separately for the X and Y directions, as shown in Figure 4. Here, as shown in Figure 2, if the filter size is S1 > S2, the Gaussian filter F0 is applied to input image IM1 such that the conversion ratio from blurred image IM11 to converted image IM2 is R2 > R1. As shown in Figure 3, if the filter size is S2 > S1, the Gaussian filter F0 is applied to input image IM1 such that the conversion ratio from blurred image IM11 to converted image IM2 is R1 > R2.

[0072] After calculating the filter sizes (S1, S2), the host device 100 sets a Gaussian filter F0 with filter coefficients KE where the standard deviation σ received on the UI screen 500 is set to standard deviations σ1 and σ2 (σ1=σ2) (S204). When the host device 100 generates a Gaussian filter F0, it calculates the value of the Gaussian distribution function f(x,y) according to the filter sizes (S1, S2) according to equation (1) above, and calculates each filter coefficient KE by normalizing the obtained values. As shown in Figure 8, if the standard deviation σi associated with the printing condition C0 is set, the host device 100 only needs to calculate the value of the Gaussian distribution function f(x,y) according to the filter sizes (S1, S2) according to equation (2) above. Alternatively, the host device 100 may pre-store multiple different Gaussian filters F0 in the storage device 114 according to the standard deviation σ=σ1=σ2, and select the Gaussian filter to which the received standard deviation σ is applied from among the multiple Gaussian filters F0 on the UI screen 500.

[0073] Subsequently, the host device 100 selects a pixel of interest PX1t from among multiple pixels PX1 that form the blurred image IM11 (S206). This process can be described as setting the coordinates of the pixel of interest PX1t in an XY coordinate plane aligned with the blurred image IM11.

[0074] Next, the host device 100 performs a filter operation by applying a Gaussian filter F0 to the input image IM1, centering on the pixel of interest PX1t (S208). Note that if the color system of the input image IM1 is RGB, the pixel value of each pixel PX1 is a combination of R, G, and B values. Here, let Kt be the S1 × S2 filter coefficients KE contained in the Gaussian filter F0, and let Pt be the pixel value of pixel PX1 in the input image IM1 that matches the filter coefficient Kt. If Q is the pixel value of pixel PX1 in the blurred image IM11, the filter operation is performed according to the following equation.

number

[0075] After calculating the pixel value Q, the host device 100 branches the process depending on whether or not there are unprocessed pixels PX1 for which the pixel value Q has not been calculated (S210). If there are unprocessed pixels PX1, the host device 100 repeats the processes S206 to S210. As a result, a pixel value Q is calculated for all pixels PX1 that will become the blurred image IM11. Once a blurred image IM11 with a pixel value Q for all pixels PX1 has been generated, the host device 100 terminates the blurred image generation process. As described above, the blurring processing unit U1 applies a Gaussian filter F0 to the input image IM1, which has filter coefficients KE such that the standard deviation σ2 matches the standard deviation σ1, even if the filter sizes (S1, S2) are different in the X and Y directions. The standard deviations σ=σ1=σ2 of the Gaussian filter F0 are the standard deviations set in the UI screen 500 shown in Figure 10. As shown in Figure 8, if a standard deviation σi is set that is linked to the printing condition C0, the blurring processing unit U1 applies the Gaussian filter F0 to the input image IM1 using the standard deviation σi linked to the printing condition C0 as standard deviations σ1 and σ2.

[0076] Furthermore, in the UI screen 500 shown in Figure 10, if the print condition C0 does not change, the standard deviation σ applied to the Gaussian filter F0 will not change even if the output resolution is changed in the resolution selection field 506. Referring to Figure 8, the blurring processing unit U1 will match the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the second output resolution RE2 is accepted to the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the first output resolution RE1 is accepted. As shown in Figure 8, if a standard deviation σi is set that is linked to the type of medium ME0, the blurring processing unit U1 will make the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the second medium candidate CM2 is accepted larger than the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the first medium candidate CM1 is accepted. As shown in Figure 8, if a standard deviation σi is set that is associated with the type of input image IM1, the blurring processing unit U1 makes the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the second image candidate C12 is accepted larger than the standard deviations σ1 and σ2 of the Gaussian filter F0 applied to the input image IM1 when the first image candidate C11 is accepted.

[0077] After the blurred image generation process is completed, the host device 100 performs a resolution conversion process to generate a converted image IM2 from the blurred image IM11, with the resolution converted to match the conversion magnification (R1, R2) calculated in S106 of Figure 9 (S112 of Figure 9). Note that since edges such as lines contained in the input image IM1 are weakened by the blurred image generation process, depending on the type of input image IM1, the host device 100 may perform a sharpening process to enhance edges between S110 and S112. If the blurred image generation process in S110 is not performed, the host device 100 performs a resolution conversion process in S112 to generate a converted image IM2 from the input image IM1, with the resolution converted to match the conversion magnification (R1, R2) calculated in S106. The host device 100 can generate the converted image IM2 from the blurred image IM11 or the input image IM1 by interpolation calculations using the bicubic method, etc., which references up to 4 × 4 pixels PX1 based on a reference point.

[0078] After generating the converted image IM2, the host device 100 processes the converted image IM2 into, for example, two C, M, Y, and K elements. 8 A color conversion process is performed to convert the data into ink amount data having integer values ​​of the gradation (S114). The color conversion process in S114 can be, for example, a process that converts the R, G, and B values ​​of each second pixel PX2 into ink amount data while referring to a color conversion lookup that defines the correspondence between the gradation values ​​of R, G, and B and the gradation values ​​of C, M, Y, and K.

[0079] Next, the host device 100 reduces the number of gradations in the gradation values ​​of each pixel constituting the ink amount data by performing a predetermined halftone process, thereby generating halftone data (S116). For the halftone process, dithering, error diffusion, density pattern, etc., can be used. The halftone data may be binary data representing the dot formation state on a pixel-by-pixel basis, indicating whether or not a dot is formed, or it may be multi-level data with three or more gradations that can handle dots of different sizes, such as small, medium, and large.

[0080] Subsequently, the host device 100 transmits the halftone data to the printer 200, causing the printer 200 to form a print image IM3 based on the converted image IM2 (S118), and then terminates the print control process. Upon receiving the halftone data, the printer 200 ejects ink droplets from the print head 220 so that the print image IM3 is formed on the medium ME0 based on the halftone data. As described above, the control unit U10, which includes the color conversion unit U3 and the halftone processing unit U4, controls the ejection of ink from the print head 220 so that a printed image IM3 based on the converted image IM2 is formed on the medium ME0.

[0081] Furthermore, if the printer 200 is capable of performing halftone processing, the host device 100 may send ink quantity data to the printer 200, and the printer 200, upon receiving the ink quantity data, may perform halftone processing. If the printer 200 is also capable of performing color conversion processing, the host device 100 may send the converted image IM2 to the printer 200, and the printer 200, upon receiving the converted image IM2, may perform color conversion processing.

[0082] When the above-described print control process is performed, a printed image IM3 is formed based on the converted image IM2 in which thin lines are properly connected, even if the resolution conversion ratios (R1, R2) differ in the X and Y directions. Furthermore, even if the output resolution RE changes under the set print condition C0, the line width of the printed image IM3 can be made uniform while obtaining an appropriate blurring effect.

[0083] As shown in Figure 2, when a Gaussian filter F0 with a filter size of S1 > S2 is applied to the input image IM1 when the conversion ratio R2 > R1, the following effect is obtained. Because the blur range in the X direction is wider than the blur range in the Y direction, the connection of thin lines L0 in the direction closer to the X direction than the Y direction is ensured. Also, because the blur range in the Y direction is narrower than the blur range in the X direction, the blurring in the Y direction is not excessive. Furthermore, in the Gaussian filter F0, the standard deviation σ2 in the Y direction matches the standard deviation σ1 in the X direction, so the blurring effect in the X and Y directions is appropriate, as is the blurring effect in the diagonal direction between the X and Y directions. If the resolution in the X direction changes while the resolution in the Y direction remains unchanged under the set printing condition C0, the blur range in the Y direction remains constant regardless of the resolution in the X direction, thus ensuring the thickness of thin lines L2 along the X direction.

[0084] As shown in Figure 3, when a Gaussian filter F0 with a filter size of S2 > S1 is applied to the input image IM1 when the conversion ratio is R1 > R2, the following effect is obtained. Because the blurring range in the Y direction is wider than the blurring range in the X direction, as shown in Figure 6, the connection of thin lines in the direction closer to the Y direction than the X direction is ensured. Also, because the blurring range in the X direction is narrower than the blurring range in the Y direction, the blurring in the X direction is not excessive. Furthermore, in the Gaussian filter F0, the standard deviation σ1 in the X direction matches the standard deviation σ2 in the Y direction, so the blurring effect in the Y and X directions is appropriate, as is the blurring effect in the diagonal direction between the Y and X directions. When the resolution in the Y direction changes while the resolution in the X direction remains unchanged under the set printing condition C0, because the blurring range in the X direction is narrower than the blurring range in the Y direction, as shown in Figure 7, the thin lines L2 along the Y direction do not become too thick and the blurring in the X direction is not excessive. As a result, the thin lines L2 along the Y direction do not become too thin. A similar effect can be obtained when the resolution in the X direction changes, but the resolution in the Y direction remains unchanged under the set print condition C0.

[0085] (4) Variations: Various modifications of this invention are conceivable. For example, the blurring processing unit U1 and the resolution conversion unit U2 may be provided in the printer 200. Therefore, the print control processing shown in Figure 9 may be performed by the printer 200. The entity performing the print control processing is not limited to the CPU; it may also be an electronic component other than the CPU, such as an ASIC. Here, ASIC is an abbreviation for Application Specific Integrated Circuit. Of course, multiple CPUs may cooperate to perform the print control processing, or a CPU and other electronic components (such as an ASIC) may cooperate to perform the print control processing.

[0086] The color system of the input image IM1 is not limited to RGB; it may also be CMY, CMYK, etc. The color systems of the blurred image IM11 and the converted image IM2 will be matched to the color system of the input image IM1.

[0087] The processes described above can be modified as needed, such as by changing the order of operations. For example, the decision process in S108 may be removed from the print control process shown in Figure 9, and the blurred image generation process in S110 may be performed by the image processing device U0 even if R1≧1 and R2≧1. The filter sizes (S1, S2) of the Gaussian filter F0 are not limited to the calculated values ​​shown in S202 in Figure 11. For example, the first filter size S1 in the X direction may be a number greater than the smallest odd number greater than the reciprocal of the first transformation magnification 1 / R1 plus 2, for example. The second filter size S2 in the Y direction may be a number greater than the smallest odd number greater than the reciprocal of the second transformation magnification 1 / R2 plus 2, for example.

[0088] In the UI screen 500 shown in Figure 10, at least a portion of the recommended standard deviation display area 503, the standard deviation adjustment area 504, and the standard deviation input area 505 may be omitted. The standard deviation σi associated with the set printing condition C0 is applied to the Gaussian filter F0 as standard deviations σ1 and σ2, thereby achieving a high-quality printed image corresponding to the printing condition C0.

[0089] Furthermore, if the filter sizes (S1, S2) differ in the X and Y directions, and the standard deviations (σ1, σ2) of the Gaussian filter F0 are the same in both directions, a fundamental effect can be obtained where the thin lines are properly connected in the converted image, even when the resolution conversion ratio differs depending on the direction. This fundamental effect can also be obtained when different standard deviations σ are applied to the Gaussian filter depending on the output resolution.

[0090] (5) Conclusion: As explained above, according to the present invention, in various embodiments, it is possible to provide a technology that can obtain a converted image in which fine lines are appropriately connected when the resolution conversion ratio differs depending on the orientation. Of course, even a technology consisting only of the constituent elements of the independent claim can obtain the basic functions and effects described above. Furthermore, configurations obtained by substituting or changing the combinations of each configuration disclosed in the above-mentioned examples, configurations obtained by substituting or changing the combinations of each configuration disclosed in the prior art and the above-mentioned examples, etc., are also possible. The present invention also includes these configurations, etc. [Explanation of Symbols]

[0091] 100…Host device, 200…Printer, 220…Print head, 500…UI screen, C0…Printing conditions, C11…First image candidate, C12…Second image candidate, CM1…First media candidate, CM2…Second media candidate, D1…First direction, D2…Second direction, F0…Gaussian filter, KE…Filter coefficient, L0,L1,L2…Lines, IM1…Input image, IM2…Converted image, IM3…Printed image, IM11…Blurred image, ME0…Media, ME1…First media, ME2…Second media, PR0…Image processing program, PR1…Print control program, PX1…Pixel, PX1t…Pixel of interest, PX2…Second pixel R1...First conversion magnification, R2...Second conversion magnification, RE...Output resolution, RE1...First output resolution, RE2...Second output resolution, S1...First filter size, S2...Second filter size, ST1...Blurring process, ST2...Resolution conversion process, ST3...Color conversion process, ST4...Halftone processing process, ST11...Printing condition reception process, ST12...Resolution reception process, SY1...Printing system, U0...Image processing device, U1...Blurring processing unit, U2...Resolution conversion unit, U3...Color conversion unit, U4...Halftone processing unit, U10...Control unit, U11...Printing condition reception unit, U12...Resolution reception unit, σ, σ1, σ2...Standard deviation.

Claims

1. An image processing apparatus capable of performing image processing to acquire a converted image whose resolution has been converted based on an input image having a plurality of pixels arranged in a first direction and a second direction intersecting the first direction, A blurring processing unit capable of generating a blurred image by applying a two-dimensional Gaussian filter, which has a first filter size in the first direction and a second filter size in the second direction, to the input image, The system comprises a resolution conversion unit capable of generating a converted image from the blurred image, wherein the resolution has been converted by a first conversion ratio, which is the resolution conversion ratio in the first direction, and a second conversion ratio, which is the resolution conversion ratio in the second direction. If the blurring processing unit makes the first filter size larger than the second filter size, the second conversion magnification is larger than the first conversion magnification. If the blurring processing unit makes the second filter size larger than the first filter size, the first conversion magnification is larger than the second conversion magnification. Let the standard deviation σ of the Gaussian filter in the first direction be denoted as standard deviation σ1. Let the standard deviation σ of the Gaussian filter in the second direction be taken as the standard deviation σ2. The blurring processing unit applies the Gaussian filter to the input image, wherein the second filter size is different from the first filter size, and the filter coefficients have such that the standard deviation σ2 matches the standard deviation σ1.

2. The blurring processing unit generates the blurred image when at least one of the first conversion magnification and the second conversion magnification is less than 1. The image processing apparatus according to claim 1, wherein the resolution conversion unit generates a converted image from the input image in which the resolution has been converted by the first conversion ratio in the first direction and the second conversion ratio in the second direction, when the first conversion ratio and the second conversion ratio are 1 or more.

3. The image processing apparatus according to claim 1, wherein the blurring processing unit sets the first filter size of the Gaussian filter to an odd number greater than the reciprocal of the first conversion ratio when the first conversion ratio is less than 1, and sets the second filter size of the Gaussian filter to an odd number greater than the reciprocal of the second conversion ratio when the second conversion ratio is less than 1.

4. An image processing apparatus according to any one of claims 1 to 3, A print head having multiple nozzles capable of dispensing liquid onto a medium, A printing system comprising: a control unit that controls the discharge of the liquid from the print head so that a printed image based on the converted image is formed on the medium.

5. In printing conditions where the resolution of the converted image can be set to one of several different output resolutions, the system further includes a resolution receiving unit that accepts the setting of the resolution of the converted image from among the multiple output resolutions. The output resolution includes a first output resolution and a second output resolution different from the first output resolution. The printing system according to claim 4, wherein the blurring processing unit matches the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the second output resolution is received by the resolution receiving unit to the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the first output resolution is received by the resolution receiving unit.

6. The system further includes a print condition receiving unit that accepts the setting of the aforementioned print conditions from among multiple candidates. The printing system according to claim 5, wherein the blurring processing unit can apply the Gaussian filter to the input image using the standard deviations linked to the printing conditions from among the standard deviations σi linked to each of the plurality of candidates as the standard deviations σ1 and σ2.

7. The aforementioned medium comprises a first medium and a second medium through which the liquid is less likely to seep than through the first medium. The plurality of candidates that can be the printing conditions include a first medium candidate for forming the print image on the first medium, and a second medium candidate for forming the print image on the second medium. The printing system according to claim 6, wherein the blurring processing unit makes the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the second media candidate is accepted as a printing condition greater than the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the first media candidate is accepted as a printing condition.

8. The plurality of candidates that can be the printing conditions include a first image candidate that prioritizes the sharpness of the line drawings included in the input image, and a second image candidate that prioritizes the gradation of the input image. The printing system according to claim 6, wherein the blurring processing unit makes the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the second image candidate is accepted as a printing condition greater than the standard deviations σ1 and σ2 of the Gaussian filter applied to the input image when the first image candidate is accepted as a printing condition.

9. The printing system according to claim 4, wherein the blurring processing unit receives an operation on a user interface screen to set the standard deviation σ of the Gaussian filter to be applied to the input image, and applies the Gaussian filter having the filter coefficients with the received standard deviation σ as the standard deviation σ1 and σ2 to the input image.

10. An image processing method for obtaining a converted image whose resolution has been converted based on an input image having a plurality of pixels arranged in a first direction and a second direction intersecting the first direction, A blurring process that generates a blurred image by applying a two-dimensional Gaussian filter, which has a first filter size in the first direction and a second filter size in the second direction, to the input image; The process includes a resolution conversion step of generating a converted image from the blurred image, in which the resolution has been converted by a first conversion ratio, which is the resolution conversion ratio in the first direction, and a second conversion ratio, which is the resolution conversion ratio in the second direction. In the blurring process described above, if the first filter size is made larger than the second filter size, the second conversion magnification is made larger than the first conversion magnification. In the blurring process described above, if the second filter size is made larger than the first filter size, the first conversion magnification is made larger than the second conversion magnification. Let the standard deviation σ of the Gaussian filter in the first direction be denoted as standard deviation σ1. Let the standard deviation σ of the Gaussian filter in the second direction be taken as the standard deviation σ2. An image processing method comprising: in the blurring step, applying the Gaussian filter to the input image, wherein, when the second filter size is different from the first filter size, the Gaussian filter has filter coefficients such that the standard deviation σ2 matches the standard deviation σ1.

Images