JP2013120951A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate pattern data in consideration of the MTF characteristic of an image input apparatus, the pattern data being used to perform resolution conversion from a low resolution image by matching the pattern data from within a data base and replacing it.SOLUTION: A high resolution image is acquired and downsampled 102 to generate a low resolution image. From the same coordinates of an inputted high resolution image 103 and a generated low resolution image 104, a signal value gradient pattern is acquired. Signal values acquired from both of the images are associated 105 and held in a database 106.

Description

  The present invention relates to an image processing apparatus and an image processing method, and relates to an image processing apparatus and an image processing method for converting an input image resolution from a low resolution to a high resolution and outputting the same. It relates to generation of pattern data.

  When printing a low resolution image acquired by an image reading device such as a scanner at a high resolution, it is necessary to convert the resolution of the image from a low resolution to a high resolution. In addition, when transferring images over a network, the resolution of the image is converted from a low resolution to a high resolution when transferring and displaying at a high resolution to reduce the network load. There is a need.

  Conventionally, many methods have been proposed as a method for converting the resolution of an input low resolution image to a high resolution. As an interpolation method often used for resolution conversion of multi-gradation images, the signal value of the grid point closest to the interpolation point to be interpolated is determined as it is as the signal value of the interpolation point. There is a neighborhood interpolation method. In addition, there is a bilinear interpolation method (Bi-Linear method) in which the signal value of the interpolation point is determined in a linear format based on the signal values of 2 × 2 pixels (4 pixels) around the interpolation point. Further, there is a cubic convolution interpolation method (Bi-Cubic method) that determines a signal value of an interpolation point from 4 × 4 pixels (16 pixels) around the interpolation point by a cubic equation.

  In addition, a pattern of a specific shape is acquired from the low-resolution image, a pattern of a high-resolution image corresponding to the signal value of the shape acquired in advance is prepared in the database, pattern matching is performed from the database, and replacement is performed. There is a method of interpolation.

  However, when resolution conversion is performed using the above-described conventional method, there are the following problems. In the nearest neighbor interpolation method, the algorithm is simple and high-speed processing is possible, but as the enlargement ratio increases, the mosaic pattern at the edge and the image quality degradation such as jaggy at the shaded area are observed. It is done. In addition, image quality degradation such as jaggy is reduced in the bilinear interpolation method and the cubic convolution interpolation method in which the interpolation point is determined based on the signal values of a plurality of surrounding pixels, compared to the nearest neighbor interpolation method. By averaging the signal values at a plurality of points, 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.

  Patent Document 1 discloses a method of performing smoothing on a contour portion using pattern matching. In the case of the method using pattern matching, it is possible to reduce the generation of a pattern to be used for problems such as jaggy and smoothing by taking into account the MTF (Modulation Transfer Function) characteristics.

JP-A-6-227048

  However, in the case of the technique disclosed in Patent Document 1, creation is not performed in consideration of the MTF characteristics of the image reading apparatus when a pattern used for pattern matching is acquired, and a pattern generation technique for the image input apparatus is used for resolution conversion. Is not optimal.

  The present invention relates to an image processing apparatus and image processing that perform generation taking into account the MTF characteristics of an image input device in generation of pattern data used when performing resolution conversion from a low-resolution image by performing matching and replacement from within a database. It aims to provide a method.

  The present invention has been made for the purpose of solving the above-described problems. An image processing apparatus according to the present invention includes an image acquisition unit (101) for inputting a high resolution image, an interpolation unit (102) for generating a low resolution image obtained by reducing the resolution of the high resolution image, Acquisition means (103) for acquiring a plurality of signal value patterns from, a signal value pattern acquired from within the low resolution image, and an acquisition means for acquiring a plurality of signal value patterns from the corresponding coordinates in the high resolution image (104), a signal value pattern acquired from within the low resolution image and a signal value pattern acquired from within the high resolution image are associated with each other, and processing means (105) is stored with the associated signal value data. An image processing apparatus comprising storage means (106).

  According to the present invention, a signal value (density or luminance) gradient is acquired at the time of pattern generation used when converting a resolution from a low resolution image to a high resolution image, and a pattern is generated in consideration of the MTF characteristics of the image input device. Can do.

6 is a flowchart illustrating a configuration of a function for creating a pattern stored in a low / high resolution correspondence pattern storage unit according to the first exemplary embodiment. FIG. 5 is an image diagram for specifically explaining a method of creating a pattern stored in a low / high resolution correspondence pattern storage unit in the first embodiment. 6 is a flowchart illustrating a configuration of a function for creating a pattern stored in a low / high resolution correspondence pattern storage unit that reduces noise from an image after resolution conversion in the first embodiment. 6 is a flowchart illustrating a configuration of a function for creating a pattern stored in a low / high resolution correspondence pattern storage unit that produces a filtering effect on an image after resolution conversion in the first embodiment. 5 is a flowchart illustrating a configuration for performing resolution conversion using a pattern stored in a low / high resolution correspondence pattern storage unit according to the first exemplary embodiment. 5 is a flowchart illustrating a configuration for performing resolution conversion using a pattern stored in a low / high resolution correspondence pattern storage unit according to the first exemplary embodiment. 10 is a flowchart illustrating a configuration of a function for creating a pattern stored in a low / high resolution correspondence pattern storage unit in the second embodiment. 10 is a flowchart illustrating a configuration for performing resolution conversion using a pattern stored in a low / high resolution correspondence pattern storage unit in the second embodiment. 10 is a flowchart illustrating a configuration of a function for creating a pattern stored in a low / high resolution correspondence pattern storage unit in the third embodiment. 12 is a flowchart illustrating a configuration for performing resolution conversion using a pattern stored in a low / high resolution correspondence pattern storage unit in the third embodiment. 12 is a flowchart illustrating a configuration for performing resolution conversion using a pattern stored in a low / high resolution correspondence pattern storage unit in the third embodiment.

  Hereinafter, an embodiment according to the present invention will be described in detail with reference to the drawings. However, the components described in this embodiment are merely examples, and are not intended to limit the scope of the present invention thereto.

[Example 1]
FIG. 1 shows a flowchart for executing image processing in Embodiment 1 of the present invention. The image processing using the database according to the present embodiment is efficient mainly in the image output apparatus such as a copying machine or in the image display apparatus using a display or the like, but also in image processing apparatuses and applications other than the above. Applicable. The software configuration of the image processing apparatus using this database will be described later.

  Referring to FIG. 1, a signal value matching pattern that can be used for processing for converting the resolution of image data input to the image processing apparatus to i-fold and j-fold resolution, and tones corresponding to the signal value matching pattern An example of a process for generating a signal value replacement pattern is shown.

  The high resolution image data input unit 101 is an input unit to which a high resolution image is input. The downsampling processing unit 102 downsamples the resolution of the image input from the high resolution image data input unit 101 to a resolution of 1 / i times horizontal and 1 / j times vertical. The downsampling processing can be performed by simple reduction by the nearest neighbor method in consideration of the processing time, but there is concern about the occurrence of interference fringes (moire) due to downsampling. For example, the image is obtained by interpolation such as cubic convolution interpolation. This is done so as not to cause a phase shift.

  When downsampling processing at 1 / i times horizontal and 1 / j times vertical is performed by the above method, scaling cannot be performed by the following resolution conversion processing, so for one input image, It is possible to create a pattern that can also be used for other resolution conversions by performing multiple times with different multiple magnifications.

  The low resolution signal value pattern acquisition unit 103 acquires N × M pixel signal values as signal value matching patterns from the low resolution image obtained by the downsampling process. The size of the signal value matching pattern to be acquired is arbitrary, but by acquiring a large signal value matching pattern, resolution conversion described later can be performed in consideration of a wider area. Note that the size of the signal value matching pattern can be changed depending on the resolution of the input high-resolution image. However, in consideration of the effect of increasing the size and the specifications of the CPU, 2 × 2 pixels or About 4 × 4 pixels is appropriate.

  Further, the high resolution signal value pattern acquisition unit 104 acquires the signal value of the pixel in a predetermined region of the high resolution image input from the high resolution image data input unit 101 as a signal value replacement pattern. The coordinates of the pixels from which the signal value replacement pattern is acquired are determined so as to include all the pixels corresponding to the pixels from which the signal value matching pattern acquired by the low resolution signal value pattern acquisition unit 103 has been acquired. Further, the pixel size of the signal value matching pattern is, for example, 2 × 2, whereas the pixel size of the signal value replacement pattern is, for example, 3 × 3.

  In the corresponding signal value integration unit 105, a plurality of signal value matching patterns and signal value replacement patterns acquired by the acquisition units 103 and 104 are associated with each other and written and held in the low / high resolution corresponding pattern storage unit 106. The

  By using, for example, an image acquired from a high-resolution scanner or the like as a high-resolution image used when creating a pattern, a learning pattern can be created that takes into account the MTF characteristics of the device when resolution conversion is performed. Using the resolution conversion method described later using a pattern that takes into account the MTF characteristics of the device, in addition to converting the resolution of the image by a low resolution scanner, it also provides high resolution for images that have been downsampled in consideration of the network load etc. It is possible to convert to an image equivalent to the image quality acquired by the scanner.

  FIG. 2 shows an example of a signal value matching pattern and a signal value replacement pattern stored in the low / high resolution correspondence pattern storage unit 106.

  FIG. 2A shows an example of a signal value matching pattern acquired from the low resolution image by the low resolution signal value pattern acquisition unit 103. The signal value matching pattern is formed by a region of horizontal N pixels and vertical M pixels, and is represented by a multi-tone signal value acquired from each pixel in this region. In the example of FIG. 2A, the acquisition pixel size is N = 2 and M = 2, and signal values of A1 to A4 pixels are acquired.

  FIG. 2B shows an example of a signal value replacement pattern acquired from the high resolution image by the high resolution signal value pattern acquisition unit 104. This signal value replacement pattern is a signal value replacement pattern for the same location as the signal value matching pattern of FIG. The signal value replacement pattern is formed by regions of horizontal (iN- (i-1)) pixels and vertical (jM- (j-1)) pixels, and multi-tone signal values acquired from the respective pixels in this region. It is represented by In the example of FIG. 2B, the acquired pixel size is for making the resolution twice horizontal and double vertical, and signal values are acquired at i = 2 and j = 2.

  In order to acquire as many signal value pattern combinations as possible, signal value matching pattern combinations are acquired from a large number of images and written to the low / high resolution correspondence pattern storage unit 106. In the case of a combination of signal value matching patterns having the same signal value, the average of the signal value replacement patterns is obtained and written.

In addition, when creating a signal value matching pattern from an 8-bit image, there are concerns that there may be a maximum of (N × M) ⁸ combinations of patterns, resulting in a matching time and a large memory capacity. In addition, the low-resolution signal value pattern acquisition unit 103 can perform quantization.

  FIG. 3 shows an example of low / high resolution compatible pattern creation processing that can simultaneously reduce the influence of noise generated by image compression during resolution conversion. In FIG. 3, the same components as those shown in FIG. In this flowchart, similarly to the configuration shown in FIG. 1, the high-resolution image data input unit 101, the downsampling unit 102, the low-resolution signal value pattern acquisition unit 103, the high-resolution signal value pattern acquisition unit 104, and the corresponding signal values An integration unit 105 and a low / high resolution correspondence pattern storage unit 106 are shown. In addition to these components, an image compression unit (adding compression noise) 301 is shown.

  Image compression such as block noise and mosquito noise occurs when image compression such as JPEG is performed when transferring low-resolution images or storing images via a network. An image compression unit 301 is provided in front of the low-resolution signal value pattern acquisition unit to add those noises, and compression noise is added only to the low-resolution image, and a low-resolution image with noise and a high-resolution image without noise are associated with the signal value integration unit. Corresponding at 105 and recorded in the low / high resolution correspondence pattern storage unit 106. By performing resolution conversion described later using a pattern including noise, it is possible to generate a high-resolution image with reduced noise when a low-resolution image with noise such as block noise or mosquito noise is input. . As described above, if image degradation such as image noise that should be removed from the low-resolution image is considered to occur, the resolution can be obtained by creating and using the pattern before acquiring the low-resolution image signal value in advance. It is possible to reduce image degradation at the same time as conversion.

  FIG. 4 shows an example of low / high resolution correspondence pattern creation processing that brings about a filtering effect such as edge enhancement on an image after resolution conversion. In FIG. 4, the same components as those shown in FIG. In this flowchart, similarly to the configuration shown in FIG. 1, the high-resolution image data input unit 101, the downsampling unit 102, the low-resolution signal value pattern acquisition unit 103, the high-resolution signal value pattern acquisition unit 104, and the corresponding signal values An integration unit 105 and a low / high resolution correspondence pattern storage unit 106 are shown. Further, in addition to these configurations, a filtering unit 401 is shown.

  When edge enhancement is required for an image, it is necessary to perform spatial filtering for each output image. However, filtering is performed in the filtering unit 401 in the previous stage of the high resolution signal value pattern acquisition unit 104 to generate a low / high resolution correspondence pattern storage unit 106 and perform resolution conversion described later, so that an image after resolution conversion is obtained. It is possible to achieve a sharpening effect.

  As described above, as an image used for creating a pattern, a low-resolution image uses a more realistic image when converting the resolution, and a high-resolution image uses a more ideal image when converting the resolution. By creating and using it for resolution conversion, it is possible to produce a restoration effect that suits the purpose.

  FIG. 5 shows a configuration of an image processing apparatus (image processing system) for using the low / high resolution correspondence pattern storage unit 106 described above. With reference to FIG. 5, an example of processing for converting the resolution of image data input to the image processing apparatus to i times horizontal and j times vertical will be described.

  The low resolution image data input unit 501 is an input unit for inputting low resolution image data to be subjected to resolution conversion processing. The low resolution image input by the low resolution image data input unit 501 is transmitted to the signal value acquisition unit 502. The signal value acquisition unit 502 acquires the signal value of each pixel of the N × M pixel image in the received low-resolution image as a signal value pattern (N and M are arbitrary natural numbers). Specifically, the signal value acquisition unit 502 acquires, as a signal value pattern, signal values of N × M pixels surrounding a target pixel that needs to be interpolated when resolution conversion is performed from a low resolution image to a high resolution image. The signal value pattern acquired by the signal value acquisition unit 502 is transmitted to the pattern matching unit 503.

  The pattern matching unit 503 searches the low / high resolution correspondence pattern storage unit 106 generated above for a signal value pattern that matches the received signal value pattern or has a similar signal value pattern. Then, the pattern matching unit 503 converts the obtained signal value matching pattern into i × horizontal and j × vertical, and converts the multi-tone signal value replacement pattern associated with the signal value matching pattern to low / high resolution. Acquired from the corresponding pattern storage unit 106. The high resolution signal value pattern acquired from the low / high resolution correspondence pattern storage unit 106 by the pattern matching unit is transmitted from the pattern matching unit 503 to the signal value replacement unit 504. The signal value received by the signal value replacement unit 504 is replaced as a high-resolution signal value as an interpolation signal value. The above flow is performed for all interpolation coordinates, and the high resolution image data output unit 505 outputs the high resolution image.

  The above software configuration of the image processing apparatus is merely an example, and the present invention is not limited to this. For example, a configuration may be provided in which processing such as image area separation and edge location extraction is performed on a low-resolution image, and the resolution conversion method is branched according to the region. FIG. 6 shows a flowchart of resolution conversion performed in a configuration in which the resolution conversion method is branched according to regions. In FIG. 6, the same components as those of the first embodiment shown in FIG. This flowchart shows a low-resolution image data input unit 501, a signal value acquisition unit 502, a pattern matching unit 503, a signal value replacement unit 504, and a high-resolution image data output unit 505 in the same manner as the configuration shown in FIG. Has been. In addition to these configurations, an edge location extraction unit 601 and an upsampling unit 602 are shown. For example, when a low resolution image has a character part, the edge part extraction unit 601 performs edge extraction and divides the background (background) part and the edge (character) part. For the background (background) portion, the upsampling unit 602 performs resolution conversion by, for example, nearest neighbor method or cubic convolution interpolation, and the edge (character) portion performs resolution conversion using pattern matching. It is also possible to change the resolution conversion method adaptively. In the above example, branching by edge extraction is shown. However, it is possible to divide by image area separation and change the resolution conversion method, and the present invention is not limited to edge extraction.

[Example 2]
In the first embodiment, a method of generating a signal value matching pattern used when converting a resolution from a low resolution image to a high resolution image from a plurality of high resolution images has been described. By generating a pattern from the entire surface of a plurality of high-resolution images, it is possible to generate highly versatile signal value matching pattern data corresponding to many low-resolution images. Instead, each of photos, characters, etc. It is difficult to generate patterns specialized for image attributes. Therefore, in the second embodiment, a configuration that enables generation of a matching pattern that separates an image for each object and enables replacement specialized for an attribute will be described.

  FIG. 7 shows a flowchart of processing in Embodiment 2 of the present invention. In FIG. 7, the same components as those of the first embodiment shown in FIG. In this flowchart, similar to the configuration shown in FIG. 1, the high-resolution image data input unit 101, the downsampling unit 102, the low-resolution signal value pattern acquisition unit 103, the high-resolution signal value pattern acquisition unit 104, and the corresponding signal value integration A unit 105 and a low / high resolution correspondence pattern storage unit 106 are shown. In addition to these configurations, an image attribute separation unit 701 and an attribute branching unit 702 are shown. The low / high resolution correspondence pattern storage unit 106 has the same function as that of the first embodiment, but is recorded separately for each attribute. In the second embodiment, the structure is divided into three types of characters, photographs, and illustrations. It is also possible to change the number of divisions as necessary.

  The image attribute separation unit 701 separates image attributes for each object based on the signal values of the target pixel and its surrounding pixels for each target pixel with respect to the image input from the high resolution image data input unit 101. As the type of separation for each object, for example, separation is performed on a character portion, a photograph portion, an illustration portion, a background (background), and the like. Based on the attribute data acquired by the image attribute separation unit 701, the signal value acquired by the low resolution signal value pattern acquisition unit 103 and the high resolution signal value pattern acquisition unit 104 is branched by the attribute branching unit 702, and a corresponding signal is provided for each object. In the value integration unit 105, the high / low resolution signal values are associated and written to the low / high resolution correspondence pattern storage units 106.

  By branching as shown above, for example, many replacement patterns with a steeper density (signal value) gradient are acquired in the character part, and replacement with a gentler density (signal value) gradient is obtained in the photograph part. When a pattern is acquired and resolution conversion is performed using another pattern according to the attribute by the resolution conversion method described later, higher-precision resolution conversion is possible.

  In the above description, the attribute is determined by the high-resolution image in the image attribute separation unit 701. However, it is also possible to perform the processing on the low-resolution image after downsampling from the viewpoint of reducing processing time.

  FIG. 8 shows an example of a resolution conversion technique performed using the low / high resolution correspondence pattern storage unit 106 created using the technique described above. In FIG. 8, the same components as those shown in FIG. This flowchart shows a low-resolution image data input unit 501, a signal value acquisition unit 502, a pattern matching unit 503, a signal value replacement unit 504, and a high-resolution image data output unit 505 in the same manner as the configuration shown in FIG. Has been. In addition to these configurations, an image attribute separation unit 701 and low / high resolution correspondence pattern storage units 106 are shown, both having the functions used in FIG.

  The low resolution image data input unit 501 is an input unit for inputting low resolution image data to be subjected to resolution conversion processing. The image attribute separation unit 701 performs image attribute determination on the image input by the low resolution image data input unit 501. In this embodiment, the character part, the photograph part, and the illustration part are divided into three types. However, the number of divisions can be changed according to the number of matching pattern data to be held. The signal value of the input low-resolution image and the attribute separation result are transmitted from the image attribute separation unit 701 to the signal value acquisition unit 502. The signal value acquisition unit 502 acquires the signal value of each pixel of the N × M pixel image in the received low-resolution image as a signal value pattern. The obtained N × M size signal value pattern is branched according to the attribute separation result, and transmitted to the pattern matching unit 503 having the low / high resolution correspondence pattern storage unit 106 corresponding to the attribute. The pattern matching unit 503 performs a search from the low / high resolution correspondence pattern storage unit 106 specialized for attributes, thereby selecting a more suitable replacement pattern and performing resolution conversion. The replacement pattern acquired from the pattern matching unit 503 is transmitted to the signal value replacement unit 504, and the signal value is replaced as an interpolated signal value as a high-resolution signal value. The above flow is performed for all interpolation coordinates, and the high resolution image data output unit 505 outputs the high resolution image.

  As described above, by performing attribute separation for each pixel on the input image, a pattern can be generated for each attribute, and a pattern can be generated according to the attribute. Further, by using a pattern for each attribute, an adaptive pattern can be selected at the time of resolution conversion, and more optimal resolution conversion can be performed.

[Example 3]
In the first and second embodiments, a method for generating a versatile pattern that does not depend on a document for which resolution conversion is performed when generating a matching pattern used for resolution conversion has been described. However, it is possible to generate a highly versatile learning pattern that can be applied to other manuscripts, but once created, it is difficult to update. Therefore, in the third embodiment, a configuration in which a user or the like can change / update the pattern database to a stronger one will be described.

  When transferring an image from a device having an image reading device such as a scanner to a copying machine via a network, the resolution is reduced from the resolution at the time of scanning in consideration of the network load. When the devices connected via the network are similar scanners or scanners having similar MTF characteristics, the same low / high resolution corresponding pattern storage unit 106 shown in the first and second embodiments can be used. The same MTF characteristics can be created and stored in a low- and high-resolution compatible pattern storage unit that can create an original pattern specialized for the scanner in the pattern creation mode of the device that receives the image. When a low-resolution image is received from a scanner having the above, a pattern considering MTF characteristics specialized for the device on the transmission side is used to enable resolution conversion with higher accuracy.

  FIG. 9 shows a flowchart showing the third embodiment of the present invention. In FIG. 9, the same components as those of the first embodiment shown in FIG. This flowchart shows a downsampling unit 102, a low resolution signal value pattern acquisition unit 103, a high resolution signal value pattern acquisition unit 104, and a corresponding signal value integration unit 105, as in the configuration shown in FIG. . In addition to these components, a scanner image input unit 901, scanner information 902, and a low / high resolution correspondence pattern accumulation storage unit 903 are shown.

  When the user gives a pattern creation signal in the pattern creation mode or the like, the chart on the scanner is scanned and an image is input from the scanner image input unit 901. The input image at that time has an optimal image resolution at the highest spec of the scanner. As for the image input from the scanner image input unit 901, a high-resolution image and a low-resolution image are generated in all pixels capable of pattern generation, as in the first embodiment, and are associated by the corresponding signal value integration unit 105. It is written in the pattern accumulation storage unit 903. However, in the first embodiment, averaging is performed for the same type of patterns found in plural, but in the third embodiment, each signal value is added and recorded cumulatively. In addition, the number of counters of the pattern added for final pattern averaging is similarly written. In the third embodiment, since it is necessary to determine which type of scanner the image to be subjected to resolution conversion was acquired, the scanner information 902 used for pattern generation is stored in the low / high resolution corresponding pattern cumulative storage unit 903. Write as header information. As described above, the low / high resolution correspondence pattern cumulative storage unit 903 simultaneously stores the cumulative value of signal values acquired from the associated low / high resolution image, the number of cumulative counters, and scanner information that can specify the scanner model. Keep writing.

  FIG. 10 shows an example of a resolution conversion technique performed using the low / high resolution correspondence pattern accumulation storage unit 803 created using the technique shown above. In FIG. 10, the same reference numerals are given to the same components as those described in the first embodiment. In this flowchart, a signal value acquisition unit 502, a pattern matching unit 503, a signal value replacement unit 504, and a high-resolution image data output unit 505 are shown, as in the configuration shown in FIG. In addition to these components, a scan image input unit 1001 and a signal value calculation unit 1002 are shown.

  In addition to the image whose resolution is to be converted, scanner information, which is scanner model information, is input from the scan image input unit 1001. For example, the image of the scanned image input unit 1001 is a low-resolution image sent over the network. When the scanner information written in the low / high resolution correspondence pattern accumulation storage unit 903 and the scanner information of the image input from the scan image input unit 1001 do not match, the flow of Example 1 shown in FIG. 5 is performed. If the scanner information matches, the third embodiment is executed, and the signal value is acquired by the signal value acquisition unit 502 as in the first embodiment, and the search by the pattern matching unit 503 is performed. The pattern matching unit 503 searches for and acquires a replacement pattern from the low / high resolution correspondence pattern accumulation storage unit 903. The acquired signal value pattern is transmitted to the signal value calculation unit 1002, and the pattern is averaged using the cumulative value of the signal value acquired from the low / high resolution image and the cumulative counter number. The averaged image is replaced by the signal value replacement unit 504 as an interpolation value of the high resolution image, and after processing all the interpolation pixels, the high resolution image data output unit 505 outputs the high resolution image.

  By performing as described above, an original pattern specialized for the scanner can be created by a user or the like. In this embodiment, the pattern is used in the pattern creation mode. However, the present invention can be applied to normal copying and the like, and is not limited.

  In this embodiment, an example of transfer via a network has been described. However, the present invention can also be effectively used in the case of a copying machine holding a hard disk. When scanning an image and storing it on the hard disk, the capacity increases if the image quality is maintained as it is. However, using this example, the resolution can be reduced and the original pattern can be stored before printing or transfer to a computer. Higher resolution is also possible.

  FIG. 11 shows an example of resolution conversion technique in the case of a copying machine holding a hard disk. In FIG. 11, the same components as those shown in FIG. In this flowchart, similarly to the configuration shown in FIG. 10, a scanned image input unit 1001, a signal value acquisition unit 502, a pattern matching unit 503, a signal value replacement unit 504, a high-resolution image data output unit 505, a signal value calculation unit 1002 and a low / high resolution correspondence pattern accumulation storage unit 903 are shown. In addition to these components, a downsampling unit 1101 and a hard disk 1102 are shown.

  When the scanned image is stored in the hard disk in the copying machine, the resolution is reduced in the downsampling unit 1101 and stored in the hard disk 1102 in order to reduce the hard disk load. The downsampling unit 1101 reduces the resolution by, for example, cubic convolution interpolation and saves it. When printing an image whose resolution has been reduced and stored on the hard disk 1102, it is not necessary to check because the scanned scanner information is already known. With regard to the resolution conversion method, processing equivalent to the method shown in FIG. 10 can be performed, and resolution conversion suitable for the scanner can be performed.

101 High-resolution image data input unit 102 Downsampling processing unit

Claims (5)

Image acquisition means (101) for inputting a high-resolution image;
Interpolation means (102) for generating a low resolution image obtained by reducing the resolution of the high resolution image;
Obtaining means (103) for obtaining a plurality of signal value patterns from within the low resolution image;
An acquisition means (104) for acquiring a plurality of signal value patterns from the same coordinate in the high-resolution image and the signal value pattern acquired from the low-resolution image;
When the signal value pattern acquired from within the low-resolution image is associated with the signal value pattern acquired from within the high-resolution image, processing means (105);
Storage means (106) for storing the associated signal value data;
An image processing apparatus comprising: An image area separation means (601) for separating a high-resolution image or a low-resolution image obtained by reducing the resolution of a high-resolution image for each object with an attribute in the image;
Branching means (602) for branching according to the separation result by the image area separation;
The image processing apparatus according to claim 1, further comprising: Image acquisition means (101) for inputting a high-resolution image;
Interpolation means (102) for generating a low resolution image obtained by reducing the resolution of the high resolution image;
Obtaining means (103) for obtaining a plurality of signal value patterns from within the low resolution image;
An acquisition means (104) for acquiring a plurality of signal value patterns from the same coordinate in the high-resolution image and the signal value pattern acquired from the low-resolution image;
When the signal value pattern acquired from within the low-resolution image is associated with the signal value pattern acquired from within the high-resolution image, processing means (105);
Storage means (106) for storing the associated signal value data;
Acquisition means (802) for acquiring image input device information for acquiring a high-resolution image;
The image processing apparatus according to claim 1, further comprising: Noise adding means (301) for adding noise to an image from a high resolution image inputted to a low resolution image;
The image processing apparatus according to claim 1, further comprising: Filtering means (401) for filtering the input high resolution image;
The image processing apparatus according to claim 1, further comprising: