JP6157234B2 - Image processing apparatus, image processing method, and program - Google Patents

Description

  The present invention relates to a resolution reduction process for image data.

  When printing low resolution image data acquired by an image reading device such as a scanner at high resolution, it is necessary to convert the resolution of the image data from low resolution to high resolution. In addition, when transferring image data via a network, when high-resolution image data is transferred at a lower resolution in consideration of the network load, it is necessary to display and print after returning to high resolution at the network destination. is there.

  Conventionally, many methods have been proposed as a method for converting an input low resolution image into a high resolution. For example, the nearest neighbor method (Nearest Neighbor method) is an interpolation method often used when performing resolution conversion on multi-gradation images. In this method, the signal value is determined as it is as the signal value of the interpolation point. Further, in the bilinear interpolation method (Bi-Linear method), the signal value of the interpolation point is determined by a linear format based on the signal value of the lattice points of 2 × 2 pixels (4 pixels) around the interpolation point. Further, in the cubic convolution interpolation method (Bi-Cubic method), the signal value of the interpolation point is determined by a cubic equation from 4 × 4 pixels (16 pixels) around the interpolation point.

JP-A-6-227048

  When resolution conversion is performed using the above conventional method, there are the following drawbacks.

  First, in the case of nearest neighbor interpolation, the algorithm is simple and it is possible to perform processing at high speed.However, as the enlargement ratio increases, a mosaic pattern is formed at the edge portion, and a jaggy or the like is formed at the shaded portion. Degradation of image quality is observed.

  In addition, in the case of bilinear interpolation or cubic convolution interpolation in which interpolation points are determined based on signal values of a plurality of neighboring pixels, image quality degradation such as jaggy is reduced compared to nearest neighbor interpolation. However, when the signal values at a plurality of points are averaged, the image is smoothed. Therefore, rounding becomes conspicuous in a portion where sharp image quality is required, such as an edge portion of an image or characters. Further, when an image whose resolution has been reduced once is increased in resolution, a portion that needs to be restored in detail, such as small-point characters, cannot be finely restored.

  Furthermore, a method of performing smoothing on the contour portion using pattern matching has been proposed (see Patent Document 1). In this case, it is possible to reduce the generation of a pattern to be used for problems such as jaggies and smoothing by taking into account MTF (Modulation Transfer Function) characteristics. However, even in the case of the technique disclosed in Patent Document 1, when an image once reduced in resolution is increased in resolution, a portion that needs to be restored in detail, such as small-point characters, cannot be finely restored.

An apparatus according to the present invention is an apparatus for converting the resolution of an input image to a low resolution, and includes a target pixel in the input image and a phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels. An image after resolution reduction by integrating the tone number conversion means for reducing the tone number of the signal value and the signal value of the target pixel having the reduced tone number and the signal value of the phase shift pixel having the reduced tone number Signal value integration means for determining signal values of the pixels corresponding to the target pixel and the phase-shifted pixel, and the predetermined number of pixels is determined according to the degree of resolution reduction. Features.

  According to the low resolution processing according to the present invention, when a reduced resolution image is later returned to a high resolution image, the detailed portion can be accurately restored.

1 is a diagram illustrating an example of a system configuration including an image forming apparatus that performs resolution reduction processing according to Embodiment 1. FIG. (A) is a diagram showing the hardware configuration of the MFP, (b) is a diagram showing the hardware configuration of the printer, and (c) is a diagram showing the hardware configuration of the cloud server. 3 is a flowchart showing a rough processing flow from acquisition of image data by an MFP to print output of image data by a printer in the system configuration of the first embodiment. FIG. 3 is a block diagram illustrating an internal configuration of a resolution conversion unit according to the first embodiment. 6 is a flowchart illustrating a flow of a resolution reduction process according to the first embodiment. It is a figure explaining the mode of the gradation number conversion process in the case of converting the resolution of an image into the resolution of 1/2 horizontal and 1/2 vertical. It is a figure explaining the process in which the signal value of the corresponding pixel in the image after resolution reduction is determined. It is a figure explaining a mode that the acquired signal value is divided | segmented as a signal value of the new attention pixel and new phase shift | offset | difference pixel in the image after high resolution conversion. It is a figure which shows how the signal value into which the new attention pixel and the new phase shift pixel were divided | segmented is rearranged to an object coordinate. It is a figure which shows the detail of a signal value gradient pattern matching part. It is a figure explaining the mode of a pattern matching process. FIG. 10 is a block diagram illustrating an internal configuration of a low resolution processing unit in a resolution conversion unit according to a second embodiment. 10 is a flowchart illustrating a flow of a resolution reduction process in a low resolution processing unit according to the second embodiment. FIG. 6 is a diagram illustrating an outline of Example 2. FIG. 10 is a block diagram illustrating an internal configuration of a low resolution processing unit in a resolution conversion unit according to a third embodiment. 14 is a flowchart illustrating a flow of a resolution reduction process in a low resolution processing unit according to the third embodiment. It is a flowchart which shows the detail of an auxiliary information embedding determination process. FIG. 10 is a diagram for explaining a resolution reduction process in Embodiment 3, where (a) shows an image before the resolution reduction process, and (b) shows a part of the image after the resolution reduction process.

  Hereinafter, embodiments according to the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram illustrating an example of a system configuration including an image forming apparatus that performs resolution reduction processing according to the present embodiment. A system 100 shown in FIG. 1 includes an image forming apparatus (MFP) 101 having a scanner function, an information processing apparatus (cloud server) 102 having an image processing module, and an image forming apparatus (printer) 103 having no scanner function. Configured and connected to each other via a network 104. In the example of FIG. 1, image data acquired by the scanner of the MFP 101 is sent to the cloud server 102, and image data transferred through image processing such as color processing in the cloud server 102 is sent to the printer 103 and output. It shows how it works.

  2A is a diagram illustrating a hardware configuration of the MFP 101, FIG. 2B is a diagram illustrating a hardware configuration of the printer 103, and FIG. 2C is a diagram illustrating a hardware configuration of the cloud server 102.

  In FIG. 2A, a CPU 201 is a processor that comprehensively controls the MFP 101, and controls each unit connected via the internal bus 209. The display unit 202 includes a liquid crystal panel having a touch screen function and displays various information, and the user gives a scan instruction via a screen displayed on the display unit 202. The scanner unit 203 has a function of scanning a document and acquiring image data. The printer unit 204 has a function of printing an image on a recording medium such as paper. The memory 205 includes a ROM that stores various commands (including application programs) executed by the CPU 201 to control the MFP 101 and various data, and a RAM that functions as a work area for the CPU 201. The HDD 206 temporarily stores image data scanned by the scanner unit 203. The network interface 207 transmits and receives image data and the like via the network 104 according to the CPU 201. The resolution conversion unit 208 performs processing for converting the resolution of various images. Details of the resolution conversion unit 208 will be described later.

  In FIG. 2B, a CPU 211 is a processor that comprehensively controls the printer 103, and controls each unit connected via the internal bus 218. The display unit 212 is composed of a liquid crystal panel or the like, and displays various information. The printer unit 213 has a function of printing an image on a recording medium such as paper. The memory 214 includes a ROM that stores various commands (including application programs) executed by the CPU 211 to control the printer 103 and various data, and a RAM that functions as a work area for the CPU 211. The HDD 215 temporarily stores received image data and the like. The network interface 216 transmits and receives image data and the like via the network 104 according to the CPU 211. The resolution conversion unit 217 performs processing for converting the resolution of various images. Details of the resolution conversion unit 217 will be described later.

  In FIG. 2C, a CPU 221 is a processor that comprehensively controls the cloud server 102 and controls each unit connected via the internal bus 227. The memory 222 includes a ROM that stores various commands executed by the CPU 221 to control the cloud server 102 and a RAM that functions as a work area for the CPU 221. The keyboard controller 223 controls key input from the keyboard 230. The display controller 224 controls display on the display 240. The disk controller 225 controls data access in an external memory 250 (hard disk drive (HDD or the like)) that stores various data, and the network interface 226 transmits and receives image data and the like via the network 104 in accordance with the CPU 221.

  It goes without saying that the hardware configurations of the MFP 101, the cloud server 102, and the printer 103 described above are merely examples, and are not limited to the above configurations.

  FIG. 3 is a flowchart showing a rough process flow from acquisition of image data in the MFP 101 to print output of image data in the printer 103 in the system configuration shown in FIG.

  In step 301, the MFP 101 acquires input image data. For example, captured image data captured by a camera and raster image data created by a personal computer or the like are acquired via a portable recording medium such as a USB memory or the network 104.

  In step 302, the MFP 101 performs resolution conversion processing in consideration of reduction of network load, memory load, and the like on the acquired input image data in the resolution conversion unit 208, and generates a low resolution image with a reduced image resolution. . In this resolution conversion processing (resolution reduction processing), auxiliary information embedding processing used when returning a low resolution image to a high resolution image is also performed. Details of the resolution reduction processing will be described later.

  In step 303, the MFP 101 transmits the image data generated by the resolution reduction processing (hereinafter referred to as “low resolution image data”) to the cloud server 102 via the network interface 207.

  In step 304, the cloud server 102 performs various image processing on the received low-resolution image data.

  In step 305, the cloud server 102 transmits low-resolution image data subjected to predetermined image processing to the printer 103.

  In step 306, the printer 103 performs resolution conversion processing (high resolution processing) using the auxiliary information on the received low resolution image data by the resolution conversion unit 217 to generate a high resolution image.

In step 307, the printer 103 prints out the image data generated by the high resolution processing (hereinafter referred to as “high resolution image data”).
FIG. 4 is a block diagram illustrating an internal configuration of the resolution conversion unit 208/217 according to the present embodiment. The resolution converter 208/217 includes a low resolution processor 410 and a high resolution processor 420.

<Low resolution processing>
First, the low resolution processing unit 410 will be described. The low resolution processing unit 410 includes a gradation number conversion unit 411 and a signal value integration unit 412. The low resolution processing unit 410 performs processing for reducing the resolution of an image in order to reduce the size of input image data.

  The gradation number conversion unit 411 performs gradation number conversion processing for reducing the number of gradations (bit number) of the input image data. For example, when each pixel in the input image data has an 8-bit signal value, a process of converting the pixel value into a 4-bit signal value is performed. This gradation number conversion process is not performed on all pixels of an image, but on a pixel of interest and a pixel whose phase is shifted from the pixel of interest by a predetermined number of pixels (hereinafter referred to as “phase-shifted pixel”). Only do. For example, when converting the resolution of an image to a resolution of 1/2 times horizontal and 1/2 times vertical, the target pixel and the pixel shifted from the target pixel by one pixel each in the horizontal and vertical directions. A gradation number conversion process is performed on a certain pixel. If the resolution of the image is to be converted to 1/3 horizontal and 1/3 vertical resolution, the pixel of interest is in a position that is shifted in phase by 2 pixels in the horizontal and vertical directions from the pixel of interest. The gradation number conversion process is performed on the pixel. That is, the phase shift amount from the target pixel in the phase shift pixel is determined according to the degree of resolution reduction of the input image data.

  The signal value integration unit 412 integrates and embeds the signal values of each pixel of interest and phase-shifted pixels that have been converted into 4 bits into corresponding pixels (hereinafter referred to as “corresponding pixels”) in the reduced-resolution image. Process.

  FIG. 5 is a flowchart showing the flow of the resolution reduction processing in the low resolution processing unit 410. This series of processing is performed by reading a computer-executable program describing the following procedure from the HDD in the MFP 101 or the printer 103 into a memory and then executing the program by the CPU. Here, a case will be described as an example in which the resolution of the input image is converted to 1/2 horizontal and 1/2 vertical in the MFP 101. The number of gradations of the input image is 8 bits, and the same number of gradations of the output image is maintained.

  In step 501, the low resolution processing unit 410 acquires input image data (high resolution image data). The acquired input image data is sent to the tone number conversion unit 411.

  In step 502, the gradation number conversion unit 411 determines a target pixel and a phase shift pixel in the image indicated by the acquired input image data. FIG. 6 is a diagram for explaining the state of tone number conversion processing when the resolution of an image is converted to a resolution of ½ times horizontal and ½ times vertical. FIG. It is the figure which expanded the part of the character "A" in resolution image. In FIG. 6A, first, a pixel 601 is determined as a pixel of interest, and a pixel 602 that is out of phase with the pixel 601 in the X-axis direction and the Y-axis direction is determined as a phase-shifted pixel. Has been.

  In step 503, the gradation number conversion unit 411 performs processing to acquire the signal values of the determined target pixel and phase shift pixel and reduce the gradation number (bit number). Specifically, the number of bits of the signal value of the target pixel and the phase shift pixel is converted to 1 / N (N is a natural number of 2 or more). For example, when the number of gradations of the input image is 24 bits, it is assumed that the number of gradations of the output image is 24 bits, 16 bits, and 8 bits. At this time, the values of N are 2, 3, and 4, respectively. In this embodiment, since the number of gradations of the input image and the output image is the same 8 bits, N = 2, and the number of gradations of the signal value of the pixel 601 that is the target pixel and the signal value of the pixel 602 that is the phase shift pixel is 8 bits, respectively. To 4 bits. FIG. 6B shows a pixel 601 ′ after the 4-bit conversion of the pixel of interest 601 and a pixel 602 ′ after the 4-bit conversion of the phase shift pixel 602. The signal values of the target pixel and the phase shift pixel with the reduced number of gradations are sent to the signal value integration unit 412.

  In step 504, the signal value integration unit 412 uses the signal values of the target pixel and the phase shift pixel with the reduced number of gradations, and calls the corresponding pixel in the image after the resolution reduction (hereinafter referred to as “corresponding pixel”). ) Is determined. Specifically, processing is performed to integrate and embed signal values of the target pixel and the phase-shifted pixel with the reduced number of gradations (here, 4 bits each) as upper bits and lower bits in the corresponding pixels. In the case of the nearest neighbor interpolation method, which is a conventional method, the signal value of the corresponding pixel in the image after resolution reduction is determined based on the nearest pixel. Therefore, in the example shown in FIG. 6, the 8-bit value of the pixel of interest 601 is held as it is (see FIG. 6C). On the other hand, in the method according to the present embodiment, the 8-bit signal value obtained by integrating the 4-bit signal value of the pixel of interest 601 ′ and the 4-bit signal value of the phase shift pixel 602 ′ is a corresponding pixel (pixel 604). ) (See FIG. 6D). In this case, the lower 4-bit signal value in the phase-shifted pixel 602 'is the above-described auxiliary information. FIG. 7 is a diagram illustrating a process in which the signal value of the corresponding pixel 604 shown in (d) of FIG. 6 is determined. First, the upper 4-bit signal values (“1010” and “1101” in the example of FIG. 7) are acquired for each 8-bit target pixel 601 and its phase shift pixel 602 that are input. Next, among the acquired upper 4-bit signal values, the upper 4 bits “1010” of the target pixel 601 are input to the upper 4 bits of the corresponding pixel 604, and the upper 4 bits “1110” of the phase shift pixel 602 are input to the lower 4 bits. The 8-bit signal value “10101101” obtained in this way is held as the signal value in the corresponding pixel 604 of the image after resolution reduction.

  In step 505, the low resolution processing unit 410 determines whether there is an unprocessed pixel in the image indicated by the input image data. If there is an unprocessed pixel, the process returns to step 502, and the next pixel of interest and phase shift pixel are determined. If there is no unprocessed pixel, the present process is terminated.

  As described above, the resolution of the input image is converted to, for example, a resolution of 1/2 times horizontal and 1/2 times vertical.

  As described above, the upper bit (upper 4 bits in this embodiment) of the signal value of each corresponding pixel in the low resolution image data generated by the resolution reduction processing according to this method is the signal value of the nearest pixel (in FIG. 6). 601 ′). For this reason, an image equivalent to an image subjected to resolution conversion by the nearest neighbor interpolation method is obtained.

  The conversion rate (1/2 times) and the number of gradations of input / output images (8 bits) described above are merely examples, and needless to say, the invention is not limited to these.

In this embodiment, the nearest neighbor value is used as the signal value of the target pixel and the phase shift pixel, but it is also possible to consider the peripheral pixel that is lost after the reduction, for example, by weighted averaging of the peripheral pixels. It is. This will be described with reference to FIGS. 6 (e) and (f). Assume that the pixel a is a pixel of interest, the signal values of the pixels a, c, and d are 255 and the signal value of the pixel b is 30. In the above-described method, only the signal values of the pixel a and the pixel d are considered, and the signal value of the pixel b is not considered. For this reason, the signal value 30 of the pixel b that existed in the high-resolution image that is the original image is completely lost. Therefore, the value of the nearest neighbor pixel is not used as it is as the signal value of the target pixel and the phase shift pixel, but a weighted average of the peripheral pixels is performed. For example, the value of the target pixel is determined by performing a moving average of the target pixel and eight neighboring pixels using the following formula.
(X ′, y ′) = ((x−1, y−1) + (x, y−1) + (x + 1, y−1) + (x−1, y) + (x, y) + ( x + 1, y) + (x-1, y + 1) + (x, y + 1) + (x + 1, y + 1)) / 9

In addition, the value of the target pixel may be determined by performing a moving average of the target pixel and four neighboring pixels using the following formula.
(X ′, y ′) = ((x, y−1) + (x−1, y) + (x, y) + (x + 1, y) + (x, y + 1)) / 5

Further, considering that the image is blurred due to the moving average, the pixel of interest may be weighted using the following equation.
(X ′, y ′) = ((x, y−1) + (x−1, y) + (x, y) * 2 + (x + 1, y) + (x, y + 1)) / 6

  By performing the above-described resolution reduction processing after performing such a weighted average on all the pixels, it is possible to consider the signal values of the pixels that are lost by the resolution reduction processing.

<High resolution processing>
Next, the high resolution processing unit 420 will be described.

  The high resolution processing unit 420 includes a signal value acquisition unit 421, a signal value division unit 422, a divided signal value rearrangement unit 423, and a signal value gradient pattern matching unit 424, and performs a process of increasing the resolution of an image. Here, an example will be described in which the resolution of an image converted to 1/2 horizontal and 1/2 vertical described with reference to FIGS. 6 and 7 is converted to horizontal and vertical twice. And Note that, as with the low resolution processing unit 410, the conversion rate (doubled) and the number of gradations (8 bits) described here are merely examples, and it goes without saying that the present invention is not limited to these.

  The signal value acquisition unit 421 acquires the signal value (8 bits in this embodiment) of the pixel of interest in the input image (low-resolution image).

  The signal value dividing unit 422 performs a process of dividing the signal value of the target pixel acquired by the signal value acquiring unit 411 as the signal value of the new target pixel N and the new phase shift pixel S in the image after high resolution conversion. FIG. 8 is a diagram for explaining how the acquired 8-bit signal value is divided as the signal value of the new target pixel N and the new phase shift pixel S in the image after high resolution conversion. Here, it is assumed that the signal value acquisition unit 421 acquires the 8-bit signal value “10101101” of the pixel 604 illustrated in FIG. In this case, of the acquired 8-bit signal value “10101101”, the upper 4 bits “1010” is the signal value of the target pixel 601 ′, and the lower 4 bits “1101” is the signal value of the phase shift pixel 602 ′. Therefore, of the acquired 8-bit signal value, the upper 4 bits are divided into the upper 4 bits of the new target pixel N, and the lower 4 bits are divided into the upper 4 bits of the new phase shift pixel S.

  The divided signal value rearrangement unit 423 converts the upper 4 bits of the new pixel of interest N and the upper 4 bits of the new phase shift pixel S divided by the signal value dividing unit 422 into the target coordinates of the image after resolution conversion (after high resolution). Rearrange. FIG. 9 is a diagram illustrating how the signal values (divided signal values) obtained by dividing the new target pixel N and the new phase shift pixel S are rearranged at the target coordinates. In FIG. 9, blank portions indicate pixels whose signal values are unknown.

  The signal value gradient pattern matching unit 424 determines the signal value of a pixel whose signal value is unknown based on the gradient of the divided signal values of the new pixel of interest N and the new phase shift pixel S rearranged by the divided signal value rearrangement unit 423. The process of determining is performed. FIG. 10 is a diagram illustrating details of the signal value gradient pattern matching unit 424. The signal value gradient pattern matching unit 424 includes a peripheral pixel signal value acquisition unit 1001, a pattern matching unit 1002, a low / high resolution correspondence signal value pattern table 1003, and a signal value replacement unit 1004.

  The peripheral pixel signal value acquisition unit 1001 acquires the image data in which the division signal value is rearranged by the division signal value rearrangement unit 423, and the signal value of each pixel in a predetermined region in the acquired image is obtained as a signal value pattern. Get as. Specifically, when converting from a low-resolution image to a high-resolution image, the signal value of an area (for example, 3 × 3 pixels) surrounding a pixel (see FIG. 9) whose signal value needs to be interpolated is unknown. Obtained as a resolution signal value pattern. The acquired low resolution signal value pattern is sent to the pattern matching unit 1002.

  The pattern matching unit 1002 first prepares a signal value pattern (hereinafter referred to as a “matching pattern”) having a signal value pattern that matches or is closest to the received low resolution signal value pattern. Retrieve from the value pattern list. Then, the multi-tone signal value replacement pattern (high resolution signal value pattern) associated with the obtained matching pattern is retrieved from the high resolution signal value pattern list and acquired. FIG. 11 is a diagram for explaining the pattern matching process. In FIG. 11, reference numeral 1101 denotes a 3 × 3 pixel surrounding four pixels having unknown signal values in addition to four pixels of ABCD having a signal value of the upper 4 bits of the new target pixel N and the upper 4 bits of the new phase shift pixel S. This is a low resolution signal value pattern. The signal value of the unknown pixel i at the center of the low resolution signal value pattern 1101 is obtained by the high resolution signal value pattern 1102 obtained from the high resolution signal value pattern list. The high resolution signal value pattern list is obtained when the low resolution signal value pattern is A (X, Y-1), B (X-1, Y), C (X + 1, Y), D (X, Y + 1). A high resolution signal value pattern of A (X, Y-1), B (X-1, Y), C (X + 1, Y), D (X, Y + 1), E (X, Y) corresponding to the combination. Composed. In FIG. 11, when the low resolution signal value pattern 1103 in which the luminance signal values are A = 11110000 (240), B = 11010000 (208), C = 10000000 (64), and D = 11000000 (48) is acquired. An example is shown. A pattern that matches the low resolution signal value pattern 1103 is acquired from the low resolution signal value pattern list, and a corresponding high resolution signal value pattern is acquired from the high resolution signal value pattern list. As a result, the low resolution signal value pattern 1103 is replaced with a high resolution signal value pattern 1104 having E = 10100000 (80) as the signal value of the unknown pixel i.

  By increasing the number of signal value patterns in the low resolution signal value pattern list and the high resolution signal value pattern list in the low / high resolution compatible signal value pattern table, it is possible to restore more accurately and with a high number of gradations. . In the present embodiment, the case where the signal value of the unknown pixel is obtained from the signal value pattern having a cross shape (upper, lower, left, and right four vicinity) has been described. However, the present invention is not limited to this. You may have a pattern. The acquired high resolution signal value pattern is sent to the signal value replacement unit 1004.

  The signal value replacement unit 1004 performs processing for replacing the signal value of the unknown pixel i in the high-resolution image (processing for filling the unknown pixel i with the interpolation signal value) using the received high-resolution signal pattern.

  By performing the above processes for all the interpolation coordinates, the low resolution image is converted into the high resolution image.

  According to the present embodiment, the signal value of the phase shift pixel with respect to the target pixel is incorporated as auxiliary information during the resolution reduction process. Then, by using auxiliary information when returning to a high-resolution image, it is possible to restore in detail at small-point characters and edge portions.

  In the first embodiment, auxiliary information is embedded in the entire image when the resolution of the input image is converted to a low resolution. This makes it possible to restore the details of the characters, but on the other hand, there is a possibility that the pseudo contour will be conspicuous in a portion that requires a large number of gradations such as a photograph. Next, when an object having a plurality of attributes such as characters, photographs, and graphics is included in the image, an attribute determination process such as image area separation is performed, and the mode for switching the auxiliary information embedding process for each attribute is switched. This will be described as Example 2. In addition, about the part which is common in Example 1, description is simplified or abbreviate | omitted and it shall explain centering on difference.

  FIG. 12 is a block diagram illustrating an internal configuration of the low resolution processing unit 1200 in the resolution conversion unit 208/217 according to the present embodiment. Compared to the low resolution processing unit 410 according to the first embodiment, an image area separation / determination unit 1201 and a general low resolution conversion unit 1202 are added.

  The image area separation / determination unit 1201 determines, for each object in the input image, an attribute such as a character, a photograph, or a graphic, and determines whether to perform a process of embedding auxiliary information (phase-shifted pixel signal value). I do.

  The general low resolution conversion unit 1202 performs nearest neighbor interpolation, bilinear interpolation, and cubic convolution on an object that is determined not to embed auxiliary information as a result of determination in the image region separation / determination unit 1201. A resolution reduction process is performed by a general method such as interpolation.

  FIG. 13 is a flowchart showing the flow of the resolution reduction processing in the low resolution processing unit 1200 according to the present embodiment. This series of processing is performed by reading a computer-executable program describing the following procedure from the HDD in the MFP 101 or the printer 103 into a memory and then executing the program by the CPU.

  In step 1301, the low resolution processing unit 1200 acquires input image data. The acquired input image data is sent to the image area separation / determination unit 1201.

  In step 1302, the image area separation / determination unit 1201 performs image area separation processing on the input image data, and separates into objects for each attribute such as characters, photographs, and graphics.

  In step 1303, the image area separation / determination unit 1201 determines whether or not auxiliary information described in the first embodiment is to be embedded in an object to be processed among the separated objects. In this embodiment, auxiliary information is embedded only when the object attribute is a character. If it is determined that the attribute of the object to be processed is a character, the process proceeds to step 1303 to embed auxiliary information. On the other hand, if it is determined that the object has an attribute other than characters, such as a photograph or graphic, the process proceeds to step 1308 to perform a general resolution reduction process. In this embodiment, the auxiliary information is embedded only when the object attribute is a character. However, it is possible to arbitrarily set the attribute of the object for which the auxiliary information is embedded.

  Since each process of step 1304 to step 1307 is the same as steps 502 to 505 in the flowchart of FIG. 5 in the first embodiment, description thereof will be omitted.

  In step 1308, the general low resolution conversion unit 1202 reduces the object having an attribute other than characters, such as a photograph and a graphic, by a general method such as nearest neighbor interpolation, bilinear interpolation, and cubic convolution interpolation. Perform resolution processing.

  In step 1309, the low resolution processing unit 1200 determines whether there is an unprocessed object. If there is an unprocessed object, the process returns to step 1303 to determine whether auxiliary information is embedded in the next object to be processed. On the other hand, if there is no unprocessed object, this process is terminated.

  As described above, the resolution reduction process according to the attribute of the object included in the input image is executed.

  FIG. 14 is a diagram for explaining the outline of the present embodiment.

  FIG. 14A shows an input image before the resolution reduction process is performed. FIG. 14B shows a state where the image area separation / attribute determination process is performed on the input image shown in FIG. FIG. 14C shows the signal value (8 bits) of the pixel of interest in the input image as it is for the photographic object, and the 4-bit signal value of the phase shift pixel as auxiliary information for the character object. The embedded state is shown.

  Note that, when an image whose resolution has been reduced by the method according to the present embodiment is returned to a high-resolution image, high-resolution processing corresponding to the attribute of the object in the image whose resolution has been reduced is performed. Specifically, image area separation is performed on an image whose resolution has been reduced, and the image is separated for each object. For a character attribute object in which auxiliary information is embedded, high resolution is obtained in the same manner as in the first embodiment. Process. For objects other than the character attribute in which the auxiliary information is not embedded, the resolution enhancement processing is performed by a commonly used technique such as nearest neighbor interpolation, bilinear interpolation, or cubic convolution interpolation.

  According to the present embodiment, auxiliary information is embedded only for objects that need to be restored in detail, such as characters. Therefore, an object that requires a high gradation number such as a photograph and does not require detailed restoration has a high gradation number. It can be restored as it is.

  In the second embodiment, an aspect has been described in which image area separation / attribute determination processing is performed in the resolution reduction processing, and the presence / absence of auxiliary information embedding processing is switched according to the attribute of the object. In the case of the method according to the second embodiment, the same processing is performed for all objects after image area separation, and switching is not performed for each pixel in the same object. Therefore, for example, it is not possible to deal with a case in which detailed restoration is desired at an edge portion where detailed restoration is necessary in the photo portion. In addition, for example, there is a possibility that a pseudo contour is generated in a gradation portion that requires a high number of gradations in a character portion or an edge of a large point.

  Therefore, auxiliary information is embedded in pixels such as edges where detailed restoration is necessary and the number of gradations is not important, while pixels such as flat parts where detailed restoration is unnecessary and the number of gradations is important are embedded. An embodiment in which auxiliary information is not embedded will be described as a third embodiment. In addition, about the part which is common in Example 1 and 2, description is simplified or abbreviate | omitted and it shall explain centering on difference.

  FIG. 15 is a block diagram illustrating an internal configuration of the low resolution processing unit 1500 in the resolution conversion unit 208/217 according to the present embodiment. Compared to the low resolution processing unit 410 according to the first embodiment, an auxiliary information embedding determination unit 1501 and a general low resolution conversion unit 1502 are added. The general resolution conversion unit 1502 corresponds to the general low resolution conversion unit 1202 in the second embodiment.

  The auxiliary information embedding determination unit 1501 performs processing for determining whether to perform processing for embedding a signal value of a phase-shifted pixel as auxiliary information during the resolution reduction processing.

  The general low resolution conversion unit 1502 performs nearest neighbor interpolation, bilinear interpolation, and cubic convolution on pixels that are determined not to embed auxiliary information as a result of determination by the auxiliary information embedding determination unit 1501. A resolution reduction process is performed by a general method such as an insertion method.

  FIG. 16 is a flowchart showing the flow of the resolution reduction processing in the low resolution processing unit 1500 according to the present embodiment. This series of processing is performed by reading a computer-executable program describing the following procedure from the HDD in the MFP 101 or the printer 103 into a memory and then executing the program by the CPU.

  In step 1601, the low resolution processing unit 1500 acquires input image data. The acquired input image data is sent to the auxiliary information embedding determination unit 1501.

  In step 1602, the auxiliary information embedding determination unit 1501 determines whether or not to perform the process of embedding the signal value of the phase-shifted pixel as auxiliary information in a predetermined pixel unit. Here, the predetermined pixel unit is a minimum pixel unit in the image after the reduction in resolution. For example, in the case of converting the input image to a resolution of 1/2 horizontal and 1/2 vertical. A pixel group of 2 × 2 pixels in the input image is a predetermined pixel unit. FIG. 17 is a flowchart showing details of the auxiliary information embedding determination process.

In step 1701, the auxiliary information embedding determination unit 1501 acquires the signal value of the processing target pixel corresponding to the target pixel in a predetermined pixel unit and the signal values of the surrounding pixels. 18A and 18B are diagrams for explaining the resolution reduction processing in this embodiment. FIG. 18A shows an image before the resolution reduction processing, and FIG. 18B shows a part of the image after the resolution reduction processing. Yes. In the image of (a) of FIG. 18, if the processing target pixel corresponding to the pixel of interest is a N _1, the signal value of the _three pixels of N _2, N 3, _{N 4} is obtained as the peripheral pixel It will be.

In step 1702, the auxiliary information embedding determination unit 1501 calculates a difference between the signal value of the processing target pixel corresponding to the acquired target pixel and the signal value of each peripheral pixel. In the case of FIG. 18, signal value differences are calculated between N ₁ and N ₂ , N ₁ and N ₃ , and N ₁ and N ₄ , respectively.

  In step 1703, the auxiliary information embedding determination unit 1501 determines whether to embed auxiliary information based on the calculated difference. For example, if any of the calculated differences is greater than the threshold, the determination is made such that auxiliary information is embedded. In order to take into account the signal values of pixels (blank pixels in FIG. 18A) that will be lost in the low resolution image, it is possible to apply a 3 × 3 pixel filter in advance, for example. is there.

  Returning to the flowchart of FIG.

  In step 1603, the low-resolution processing unit 1500 proceeds to step 1604 if the target pixel is a target for embedding auxiliary information according to the determination result in the auxiliary information embedding determination unit 1501, and proceeds to step 1607 otherwise.

In step 1604, the gradation number conversion unit 411 determines the processing target pixel as a target pixel and determines the phase shift pixel. In FIG. 18A, if the processing target pixel is N ₁ , the N ₁ pixel is determined as the target pixel, and the S ₁ pixel is determined as the phase-shifted pixel.

  Since each processing of step 1605 and step 1606 is the same as steps 502 to 504 in the flowchart of FIG. 5 in the first embodiment, description thereof will be omitted.

  In step 1607, the general low resolution conversion unit 1502 performs a resolution reduction process on the processing target pixel by a general method such as a nearest neighbor interpolation method, a bilinear interpolation method, or a cubic convolution interpolation method.

  In step 1608, the low resolution processing unit 1500 determines whether there is an unprocessed pixel in the image indicated by the input image data. If there is an unprocessed pixel, the process returns to step 1602, and auxiliary information embedding determination processing is performed for the next target pixel. If there is no unprocessed pixel, the present process is terminated.

As a result of such processing, for example, an image with reduced resolution as shown in FIG. 18B is obtained. The pixels 1801 to 1803 in FIG. 18B are pixels determined to perform auxiliary information embedding processing, and the signal values of the lower 4 bits of the phase shift pixels S ₁ , S ₂ , and S ₃ are used as auxiliary information. Embedded. 1804 pixel contrast is the pixel which is determined not to perform the embedding process of the auxiliary information, the signal value of the lower 4bit of S ₄ is not embedded.

  Note that when returning an image that has been reduced in resolution by the technique according to the present embodiment to a high-resolution image, high-resolution processing corresponding to each pixel in the reduced-resolution image is performed. Specifically, resolution enhancement processing is performed on the pixels in which the auxiliary information is embedded in the resolution-reduced image by the same method as in the first embodiment. For pixels in which auxiliary information is not embedded, resolution enhancement processing is performed by commonly used techniques such as nearest neighbor interpolation, bilinear interpolation, and cubic convolution interpolation.

  According to the present embodiment, auxiliary information is embedded only for pixels that need to be restored in detail. Therefore, a flat portion or the like that requires a number of gradations that is representative of gradation and does not place much importance on detailed restoration is used for the floor. The logarithm can be maintained. As a result, it is possible to restore a detailed portion such as a small point character or an edge portion while preventing a pseudo contour or the like associated with a low gradation number in a gradation portion or a flat portion.

(Other examples)
The present invention is also realized by executing the following processing. That is, software (program) for realizing the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, etc.) of the system or apparatus reads the program. It is a process to be executed.

Claims (12)

An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
Equipped with a,
The predetermined number of pixels is determined in accordance with the degree of resolution reduction . An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
With
The gradation number converting means reduces the number of bits of the signal value of the target pixel and the phase shift pixel to 1 / N (N is a natural number of 2 or more) according to the gradation number of the image after the resolution reduction.
A device characterized by that . An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
With
The signal value integrating means integrates the signal value of the pixel of interest having a bit number of 1 / N as an upper bit and the signal value of the phase shift pixel having a bit number of 1 / N as a lower bit. Is determined as the signal value of the corresponding pixel
A device characterized by that . An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
With
The upper bit in the corresponding pixel is the signal value of the nearest pixel when converted to low resolution using nearest neighbor interpolation.
A device characterized by that . An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
Means for separating the input image for each object ;
With
Only the predetermined object in the input image is processed by the tone number conversion means and the signal value integration means.
A device characterized by that . An apparatus for converting the resolution of an input image to a low resolution,
A gradation number converting means for reducing the number of gradations of the signal value of the target pixel in the input image and the phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The signal value of the pixel corresponding to the target pixel and the phase-shifted pixel in the image after the resolution is reduced by integrating the signal value of the target pixel with the reduced number of gradations and the signal value of the phase-shifted pixel with the reduced number of gradations A signal value integrating means for determining a value;
Means for determining whether to perform processing by the gradation number converting means and the signal value integrating means on the input image in a predetermined pixel unit ;
With
The gradation number converting means and the signal value integrating means are only applied to the pixels in the predetermined pixel unit, which are determined to be processed by the gradation number converting means and the signal value integrating means by the determining means. Process by
A device characterized by that . 6. The apparatus according to claim 5 , wherein the predetermined object is an object that needs to be restored in detail.   8. The apparatus according to claim 7, wherein the object that needs to be restored in detail is a character object. The determination means acquires the signal value of the processing target pixel corresponding to the target pixel in a predetermined pixel unit and the signal value of the peripheral pixel, and the signal value of the processing target pixel corresponding to the acquired target pixel and each peripheral pixel calculates a difference between signal values of each, based on the calculated difference, according to claim 6, wherein the determining whether to process by the gradation number conversion means and said signal value integrating means Equipment. The apparatus according to claim 6 , wherein the predetermined pixel unit is a minimum pixel in an image after resolution reduction. A method for converting the resolution of an input image to a low resolution,
A step of reducing the number of gradations of a signal value of a target pixel in the input image and a phase-shifted pixel whose phase is shifted from the target pixel by a predetermined number of pixels;
The gradation number conversion means integrates the signal value of the pixel of interest with the reduced number of gradations and the signal value of the phase shift pixel with the reduced number of gradations, so that the target pixel and the phase shift in the image after the resolution reduction Determining a signal value of a pixel corresponding to the pixel;
Only including,
The method is characterized in that the upper bit in the corresponding pixel is a signal value of the nearest pixel when converted to a low resolution using nearest neighbor interpolation . The program for functioning a computer as an apparatus of any one of Claims 1 thru | or 10 .

Images