JP2009070407A - Method and device for image processing, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To acquire the same frequency response characteristics as an original image also for an image of desired resolution in creating a plurality of blurred image signals from an original image signal, creating a plurality of band-limited image signals based on these image signals, transforming each band-limited image signal based on a plurality of transformation functions to prepare a plurality of transformation image signals and adding an integrated signal obtained by integrating the respective transformation image signals, to the original image signal to carry out processing for enhancing a specific frequency component of the original image signal. <P>SOLUTION: The blurred image signals and band-limited image signals are created based on a reference low resolution image signal that can reproduce a resolution image closest to a desired magnification input from a magnification input means 5. The transformation functions are shifted to the desired magnification, and the band-limited image signals are transformed to obtain the transformation image signals. The integrated signal obtained from the transformation image signals is added to the reference low resolution image signal to obtain an intermediate processed image signal Sproc', which is enlarged or reduced to the desired magnification to obtain a processed image signal Sproc. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing method and apparatus for performing processing such as emphasizing a predetermined frequency component on an image signal, and a computer-readable recording medium storing a program for causing a computer to execute the image processing method. Is.

Conventionally, a number of image processing methods and apparatuses for improving diagnostic performance of radiographic images by performing frequency enhancement processing or dynamic range compression processing using an unsharp mask image signal (hereinafter referred to as a blurred image signal) by the present applicant. Have been proposed (Japanese Patent Laid-Open Nos. 55-163472, 55-87953, Japanese Patent Laid-Open Nos. 3-222577, 10-75395, 10-171983, etc.). For example, in the frequency enhancement process, a predetermined spatial frequency component of the original image signal is enhanced by adding the original image signal Sorg minus the blurred image signal Sus multiplied by the enhancement coefficient β to the original image signal Sorg. To do. This is expressed by the following equation (1).
Sproc = Sorg + β × (Sorg−Sus) (1)
(Sproc: frequency enhanced signal, Sorg: original image signal,
Sus: blurred image signal, β: enhancement coefficient)

Japanese Patent Application Laid-Open No. 10-75395 proposes a method for preventing the occurrence of artifacts in a frequency-enhanced signal by adjusting the frequency response characteristics of the addition signal added to the original image signal. ing. In this method, first, a plurality of blurred image signals having different sharpness, that is, different frequency response characteristics are generated, and the difference between two signals in the blurred image signal and the original image signal is obtained, thereby obtaining the original image signal. A plurality of band-limited image signals (bandpass signals) representing frequency components in a limited frequency band are created, and the band-limited image signals are converted to a desired size by different conversion functions, respectively. From the above, the added signal is created by integrating the plurality of suppressed band-limited image signals. This process can be expressed by, for example, the following formula (2).
Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2,... SusN)
Fusm (Sorg, Sus1, Sus2, ... SusN)
= F1 (Sorg-Sus1) + f2 (Sus1-Sus2) + ...
+ Fk (Susk-1-Susk) + ... + fN (SusN-1-SusN) (2)
(However, Sproc: Image signal in which high frequency components are emphasized.
Sorg: Original image signal
Susk (k = 1 to N): Blur image signal
fk (k = 1 to N): conversion function for converting each band limited image signal
β (Sorg): enhancement coefficient determined based on the original image signal)

  In these frequency enhancement processes, the frequency response characteristic of the addition signal added to the original image signal can be adjusted by changing the parameter of the conversion function for converting the band limited image signal. Therefore, a processed image signal having a desired frequency response characteristic such as prevention of artifacts can be obtained depending on the definition of each conversion function.

Furthermore, Japanese Patent Laid-Open No. 10-171983 proposes a method for preventing the occurrence of artifacts in the processed signal when the frequency enhancement process and the dynamic range compression process are performed simultaneously. In this method, as described in JP-A-10-75395, a plurality of band-limited image signals are created, and a signal related to a high-frequency component of the original image signal (high-frequency component signal) based on the band-limited image signal In addition, a signal related to the low frequency component (low frequency component signal) is obtained, and the frequency enhancement process and the dynamic range compression process are performed by adding the signal related to the high frequency component and the signal related to the low frequency component to the original image signal. It is a thing. This process can be expressed by, for example, the following formula (3).
Sproc = Sorg + β (Sorg) · Fusm (Sorg, Sus1, Sus2, ... SusN)
+ D (Sorg−Fdrc (Sorg, Sus1, Sus2,... SusN)) (3)
Fusm (Sorg, Sus1, Sus2, ... SusN)
= {Fu1 (Sorg−Sus1) + fu2 (Sus1−Sus2) +
+ Fuk (Susk-1−Susk) +... + FuN (SusN−1−SusN)}
Fdrc (Sorg, Sus1, Sus2, ... SusN)
= {Fd1 (Sorg−Sus1) + fd2 (Sus1−Sus2) +
+ Fdk (Susk-1−Susk) +... + FdN (SusN−1−SusN)}
(However, Sproc: processed image signal
Sorg: Original image signal
Susk (k = 1 to N): Blur image signal
fuk (k = 1 to N): conversion function used to obtain a high-frequency component signal
fdk (k = 1 to N): conversion function used to obtain a low frequency component signal
β (Sorg): enhancement coefficient determined based on the original image signal
D (Sorg-Fdrc): determined based on the low frequency component signal
Signal for dynamic range compression
(D is a function that converts Sorg-Fdrc)

  On the other hand, in the blurred image signal used for the frequency enhancement processing described above, pixels are thinned out by performing a predetermined filtering process at predetermined intervals on the pixels of the original image signal, and then, the predetermined number of pixels are thinned out. It is created by interpolating with the interpolation method. As this filtering process, a process of removing a high frequency component of the original image signal by a low-pass filter, specifically, a process of obtaining an average value of pixel values in the filter and a weighted average value is performed. In the filtering process performed in order to obtain a blurred image signal in the above-mentioned Japanese Patent Application Laid-Open No. 10-75395 or the like, this filtering process is further performed on a signal having a small number of pixels obtained by filtering the original image signal. Thus, a plurality of different blurred image signals are created by interpolating the image signals with a small number of pixels obtained at each stage of filtering so as to have the same number of pixels as the original image signals.

  The blurred image signal is created based on the original image signal as described above. This original image signal is an image having a predetermined pixel density obtained by reading the original image signal at a predetermined reading density by a reading device. Is a reproducible digital signal. In general, when a digitized image signal is reproduced as, for example, a print output, it is known that frequency components below a certain frequency (Nyquist frequency) determined by the pixel density are correctly reproduced. That is, since the reading density, that is, the pixel density is determined in consideration of the image quality level required at the time of reproduction, it is not always constant.

  For example, in a radiation image reading / reproducing system, a radiation image of a human body recorded on a stimulable phosphor sheet is read as a digital image signal by laser beam scanning, but the reading density, that is, the pixel density is the size of the stimulable phosphor sheet. Depending on the user setting, it can be changed to an arbitrary value.

  Here, with respect to image signals having different pixel densities (reading densities), that is, Nyquist frequencies, and different image resolutions obtained based on the signals, filtering processing by the same low-pass filter and the same interpolation method are performed. When the interpolation process is performed, the frequency characteristic of the obtained band limited image signal, specifically, the frequency band of the band limited image signal differs depending on the pixel density. Thus, for example, when one original image is read at two types of reading densities and an original image signal having two types of pixel densities is obtained, a band limited image signal is obtained using the same blurred image signal, and frequency enhancement processing or Even if the dynamic range compression processing is performed, the frequency band to be emphasized or the frequency band to be compressed differs depending on the two types of original image signals.

  For this reason, information on the pixel density of the original image signal is obtained, filter coefficients are selected from a plurality of filter coefficient lists based on the information, and the original image signal is filtered by the filter of the selected filter coefficient. A method for obtaining a blurred image signal has been proposed (Japanese Patent Laid-Open No. 10-63837). Here, for example, band-limited image signals obtained by filtering the original image signals with reading densities of 5 lines / mm and 6.7 lines / mm using the same low-pass filter have different frequency bands. According to this method, the frequency band of the band limited image signal obtained from each original image signal can be made to substantially match by performing filtering processing on each original image signal using a different low-pass filter. Therefore, it is possible to create a blurred image signal having the same frequency characteristic regardless of the pixel density, thereby creating a band-limited image signal having the same frequency characteristic, and performing desired processing such as frequency enhancement processing and dynamic range as described above. The compression process can always be performed in the same manner.

  On the other hand, there are various formats such as JPEG, GIF, and TIFF for compressing the original image signal (reducing the number of data). Recently, the image signal is hierarchically decomposed for each resolution, A format for encoding and compressing data (hierarchical data) has been proposed. More specifically, this compression format decomposes the original image signal into a plurality of resolution hierarchical image signals each having a resolution ½n times that of the original image by wavelet transform or the like, and the decomposed hierarchical images for each resolution. The signals are encoded in hierarchical order and compressed as one file.

This compression method has the following characteristics.
(1) Since the image signal is not processed for each block as in the conventional DCT method used in JPEG, artifacts such as block distortion do not occur.
(2) Since the image signal is hierarchically encoded, it is only necessary to transfer information of a resolution necessary for transferring the image signal, and efficient image transfer is possible.
(3) Since the image signal is decomposed into multiple resolutions or multiple resolutions, various image processing such as frequency enhancement processing can be performed relatively easily.
(4) Space and frequency can be decomposed simultaneously by multi-resolution analysis, orthogonal transformation is performed in a wide range for a low frequency region that greatly affects coding efficiency, and a narrow range for a high frequency region. Therefore, even if quantization noise is generated in the peripheral portion of the edge in the image, the spatial spread can be suppressed. For this reason, noise is difficult to perceive.

  In addition, a file format capable of storing a plurality of different data in one file has been proposed, such as a FlashPix file proposed by Eastman Kodak Co., but such a file of the FlashPix standard is also proposed. It is also possible to store hierarchical image signals that have been decomposed into multiple resolutions.

  In this way, by decomposing the original image signal into multiple resolutions, the original image signal can be constructed from hierarchical image signals representing a plurality of images having a resolution 1 / 2n times that of the original image. An image compressed using such a signal that has been decomposed into multiple resolutions may be restored and used at the same resolution level as the original image, or may be restored and used at an intermediate resolution level depending on the application. To do. For example, when it is necessary to reproduce a high-quality image like a printer, a high-quality image having the same resolution as the original image is restored by restoring the image based on a hierarchical image signal representing an image up to the maximum resolution. Can be obtained. In addition, when it is not necessary to reproduce an image with a higher resolution than a printer like a CRT, the image signal is restored based on a hierarchical image signal representing an image having a lower resolution than the original image, and further enlarged if necessary By reducing the size, an image suitable for the CRT resolution can be reproduced although the resolution is lower than that of the original image.

  However, when an image compressed using a signal decomposed into multiple resolutions is restored at an intermediate resolution level without being restored at the same resolution level as the original image, the resolution of the restored image is the original image However, since the resolution is different from the resolution according to the presumed reading density as described in JP-A-10-63837, the above-mentioned JP-A-10-63837 is used. It is difficult to use the method described in No. 1, and a desired conversion process is performed using a conversion function to be set for the original image signal. In this case, since the frequency response characteristics are different between the original image signal and the hierarchical image signal representing an image having a lower resolution than the original image signal, the frequency enhancement processing described above is lower than the original image in the image signal decomposed into multiple resolutions. When applied to an image signal representing a resolution image, if the conversion function corresponding to the resolution of the original image is applied to the low-resolution image, the frequency response characteristics are different from those when the frequency enhancement processing is applied to the original image signal. There is a risk of an image having On the other hand, in this case as well, in order to apply the method described in Japanese Patent Laid-Open No. 10-63837, a conversion function corresponding to each level of resolution is prepared, and a frequency enhancement process is performed by selecting the conversion function according to the resolution. It is possible. However, if a conversion function is prepared according to each resolution, the number of conversion functions becomes enormous and management of the conversion function becomes very troublesome. This is a problem that occurs not only when frequency enhancement processing is performed on an image signal obtained by decomposing an original image signal into multiple resolutions, but also when frequency enhancement processing is performed when an image having a lower resolution than the original image is reproduced. is there.

  In addition, the image is not limited to being restored at an intermediate resolution level that does not reach the same resolution level as the original image, but an image of the same size as the original image restored at the same resolution level as the original image is enlarged, and the image obtained by this enlargement is On the other hand, the same problem arises when emphasis processing or the like is performed using a conversion function to be set for the original image. For example, when an image is viewed on a CRT and a part of the image is interpreted in detail, the corresponding part of the image may be enlarged and displayed or output to a film. In this case, since the spatial resolution of the enlarged image is increased despite processing using the same conversion function, the image quality of the image subjected to enhancement processing after enlargement and the original image size are the same as usual. Unlike the image quality of images that have undergone emphasis processing, interpretation may be affected.

  Further, in the image processing apparatus that performs the frequency emphasis processing as described above, not only the original image signal read at a predetermined reading density (one index of resolution) obtained in the radiation image reading / reproducing system is used. Depending on the type of input modality, an original image signal representing an image having a different reading density may be input, that is, the resolution of the input image may be diversified. In such a case, when the above-described frequency emphasis processing is performed on an image signal representing an image having a resolution different from the reference resolution by a conversion function corresponding to a predetermined resolution (hereinafter referred to as a reference resolution), it is obtained. The obtained image may have a frequency response characteristic different from that obtained when the image signal representing the image of the standard resolution is processed. In this case, a conversion function is prepared according to the resolution, and an image having the same frequency response characteristic is obtained regardless of the resolution of the image to be processed by selecting the conversion function according to the resolution and performing frequency enhancement processing or the like. A reproducible processed image signal can be obtained. However, if the conversion functions are prepared according to the resolution, the number of the conversion functions becomes enormous, and the management of the conversion functions becomes very troublesome.

  The present invention has been made in view of the above circumstances, and is for causing a computer to execute an image processing method and apparatus that can always perform frequency enhancement processing or the like in the same manner regardless of the resolution of an image to be processed. It is an object of the present invention to provide a computer-readable recording medium on which the program is recorded.

A first image processing method according to the present invention creates a plurality of band limited image signals having different band limiting characteristics from an input image signal, and a plurality of conversions set corresponding to the band limited image signals. In an image processing method for obtaining a converted image signal by performing conversion processing based on a function and obtaining a processed image signal by performing predetermined processing on the input image signal based on the converted image signal,
A reference conversion function corresponding to each band-limited image signal for the reference resolution image, which is used when the conversion process is performed on the reference resolution image having the reference resolution, is prepared, and a processing image having a resolution different from the reference resolution is prepared. When the image signal to be represented is input, the reference conversion function is relatively moved so that the frequency band of each band limited image signal corresponding to the reference conversion function is closest to the frequency band of each band limited image signal for the processing image. The moved reference conversion function is set as a conversion function for the processing image.

  Note that the term “resolution” generally has a very broad meaning, and is used to mean, for example, sharpness or to mean the pixel density of an image (or image signal). In some cases, the resolution used in the present invention refers to the sharpness. The term image size is used to mean the total number of pixels of an image (or image signal). Here, “means something related to sharpness” means, for example, increasing or decreasing the pixel density (in this case also increasing or decreasing the image size) to increase or decrease the sharpness, or reproducing while maintaining the same pixel density (can be done) ) It means to increase or decrease the frequency range or change the signal response to increase or decrease the sharpness. Therefore, for example, when the reference resolution image is converted to a low resolution image by filtering processing using pixel thinning, the pixel density is reduced, and the image size of the low resolution image is smaller than that of the reference resolution image. On the other hand, a low-sharpness blurred image obtained by interpolating and enlarging a low-resolution image obtained by the filtering process using pixel thinning has the same pixel density as the reference resolution image and the same image size. In other words, the low-resolution image obtained by the filtering process using pixel thinning is an image having a lower (smaller) resolution than the reference resolution image and an image having a smaller image size, while the blurred image is lower than the reference resolution image. Are low resolution (small) images and the same image size.

  In the above, “a plurality of conversion functions set corresponding to each band limited image signal” means that if each band limited image signal is composed of six frequency bands, six conversion functions corresponding to each frequency band are used. It becomes a function.

  In the present invention, as the reference conversion function, various forms such as a linear function, a nonlinear function, and a constant can be used (the same applies to the second method described later).

  Further, as the “predetermined processing”, for example, frequency enhancement processing for emphasizing a specific frequency component included in the original image signal as shown in the above equation (2), or original image as shown in the above equation (3). There is a dynamic range compression process for reducing the difference between the highest density and the lowest density, that is, the high density area or the low density area, or the contrast in both the high density area and the low density area so as to narrow the dynamic range. The same applies to the above method).

  In the first image processing method according to the present invention, when setting the conversion function for the processing image by “relatively moving the reference conversion function”, the number of sets of the reference conversion function and the band-limited image signal, and the reference resolution image It is desirable to perform the relative movement based on the number of common frequency ranges of the band-limited image signal for and the band-limited image signal for the processing image, and specifically, the following may be performed. That is, the number of sets of the conversion function for the processing image and each band-limited image signal corresponding to the conversion function is set so that the resolution of the processing image is 2- (p + 1) times to 2-p times (p is a positive value). When the resolution of the processing image is within the range of 2q-1 to 2q times the reference resolution (q is a positive integer), the number is increased by q. Good. Note that the number of sets is not increased or decreased when the second resolution is within the range of 2-1 to 1 times the first resolution (corresponding to q and p being 0 in the above).

  Here, when the resolution of the processing image is in the range of 2- (p + 1) times to 2-p times the reference resolution, the number p is decreased when the resolution relationship is in the above-described state. Since the number of common frequency ranges in each band limited image signal is reduced by p, when performing enhancement processing on the processing image, enhancement of the reference resolution image in each band limited image signal is performed. This is because it is not necessary to use p reference conversion functions corresponding to a frequency band on the low frequency side that is not used for processing (rather, there is a risk of causing a problem if used).

  On the other hand, when the resolution of the processing image is in the range of 2q-1 times to 2q times the reference resolution, the band-limited image signal is a frequency band on the low frequency side used for enhancement processing for the reference resolution image. Since there is a shortage of components, there is no band-limited image signal to be used for the low-frequency conversion function simply by relative movement of the reference conversion function, and the low-frequency characteristics that affect image quality This is because the number of band-limited image signals needs to be increased by q when performing the enhancement process on the processing image because the characteristic cannot be made the same as the enhancement process. In this case, only by relatively moving the reference conversion function, the number of conversion functions for the processing image is smaller than the number of band-limited image signals to be generated. It is desirable to newly define a conversion function (a shortage) corresponding to a frequency band on the high frequency side that is not used in the enhancement processing for the reference resolution image, and as a result, the conversion function for the processing image and the conversion The number of sets of each band limited image signal corresponding to the function is increased by q.

  In the first image processing method, when the resolution of the processing image is lower than the reference resolution, a conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image is selected from the plurality of conversion functions. It can also be set as a conversion function for processing.

  Here, the “conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image” means, for example, the conversion function for each frequency band when the resolution of the processing image is 1 / 2k times the reference resolution. Of these, k is a conversion function on the low frequency band side. For example, when there are six conversion functions f1 to f6 (f1 corresponds to the highest frequency band) for the image signal representing the reference resolution image, and the resolution of the processing image is ½ of the reference resolution, the highest frequency The five conversion functions f2 to f6 other than the conversion functions corresponding to the bands are referred to. Then, among the six band limited image signals obtained from the image signal representing the processing image, five band limited image signals excluding the one on the lowest frequency side are converted by five conversion functions to obtain a converted image signal. It will be.

  In the first image processing method, when an output image having an image size different from the image size of the reference resolution image is obtained, the reference resolution image is close to (preferably closest to) the image size of the output image. The converted image obtained by multiplying the image size of the image by 22k (k is an integer) is subjected to conversion processing and predetermined processing as a processing image, and then the output image is output for an image whose image size represented by the processed image signal is 22k times It is desirable to perform a scaling process so as to be equal to the image size, that is, to perform a scaling process on the processed image signal subjected to the predetermined process to obtain a scaled image signal.

  “Conversion processing and predetermined processing using a variable-magnification image as a processing image” refers to inputting a variable-magnification image signal representing a magnification-varying image as an image signal representing the processing image, and inputting the input image signal ( This means that each band limited image signal is generated from the (magnified image signal), subjected to conversion processing based on the set conversion function, and further subjected to predetermined processing.

  “Image size is 22k times” means that the image size is 22k times by multiplying the pixel density by 2k in the main scanning direction and the sub-scanning direction, respectively. Note that k is an integer, that is, an integer of zero or positive and negative. To enlarge the image size of the reference resolution image by 22k means to enlarge (k> 0) or reduce (k <0) the scaled image. In addition to this, it means that the image size of the reference resolution image is set as it is (k = 0).

  Also, in the case of “scaling processing to be equal to the image size of the output image”, “scaling” is to enlarge (magnification> 1) or reduce (magnification> 1) the image size. As well as keeping the image size as it is (magnification = 1). For example, when the image size of the output image is 22k of the image size of the reference resolution image, the image size itself of the image represented by the processed image signal obtained by performing the predetermined processing is the same as the image size of the output image. Therefore, when “scaling processing is performed so as to be equal to the image size of the output image”, the scaling factor may be set to 1. In practice, the processed image signal is directly used as the scaling-processed image signal. (That is, it is not necessary to perform the scaling process).

The second image processing method according to the present invention creates a plurality of band limited image signals having mutually different band limiting characteristics from an input image signal, and a plurality of conversions set corresponding to each band limited image signal. In an image processing method for obtaining a converted image signal by performing conversion processing based on a function and obtaining a processed image signal by performing predetermined processing on the input image signal based on the converted image signal,
A processing image having a resolution different from the reference resolution is prepared by preparing a reference conversion function corresponding to each band-limited image signal for the reference resolution image, which is used when the conversion processing is performed on the reference resolution image having the reference resolution. When the image signal representing the image is input, the response of the processed image signal for the processing image is substantially the same as the response of the processed image signal for the reference resolution image based on the difference between the resolution of the processing image and the reference resolution. The reference conversion function is corrected so as to be the same (a new function is calculated), and the function obtained by this correction is set as a conversion function for the processing image.

  “Based on the difference between the resolution of the processing image and the reference resolution, the reference conversion function is set so that the response of the processed image signal for the processing image is substantially the same as the response of the processed image signal for the reference resolution image. When performing “correction”, for example, each band-limited image signal obtained from an image signal representing a reference resolution image and an image signal representing a processing image, a desired frequency response characteristic, and each image should be set. It is possible to use a method such as obtaining a parameter by solving the relational expression of the parameter of the conversion function as a simultaneous equation with each band-limited image signal and frequency response characteristic as a known value and using the parameter of the conversion function as a variable. Alternatively, it may be a method of calculating by performing operations such as interpolation and extrapolation from the frequency band of a plurality of image signals obtained from the conversion function for the processing image and the image signal representing the processing image.

  In the second image processing method, correction of the reference conversion function (calculation of a new function) is performed on the image represented by the processed image signal obtained by processing the image signal representing the input processing image. The characteristic is almost the same as the characteristic of the image represented by the processed image signal obtained by processing the image signal representing the reference resolution image in a frequency range at least 1/5 or less of the Nyquist frequency corresponding to the resolution of the processing image. It is desirable to do so.

  Furthermore, in the first and second image processing methods according to the present invention, it is preferable to obtain resolution information regarding the resolution of the processing image and set a conversion function for the processing image based on the resolution information.

  When the resolution information is acquired, the user (operator) may input a numerical value from the keyboard, or may have a form in which several types of resolution information are displayed on the operation screen and selected by the user, or for processing. Resolution information may be attached to an image signal representing an image, and the resolution information may be automatically acquired when the image signal representing the processing image is processed, and any resolution information can be acquired. Such a form may be sufficient.

  The “resolution information” may be any information as long as it is information relating to the sharpness of the image (or image signal). For example, the reading density at the time of reading the radiation image recorded on the stimulable phosphor sheet In addition to (for example, dpi or book / mm), for example, it may be a sampling frequency for obtaining an image signal.

  Each band-limited image signal in the present invention is generated by any method as long as it represents a predetermined frequency band component of an image signal (original image signal) representing an input processing image. For example, the original image signal can be created by multi-resolution expansion by a technique such as wavelet transform or Laplacian pyramid.

  Further, the band-limited image signal is generated based on the input image signal and a plurality of blurred image signals having different frequency response characteristics, and is generated based on the plurality of blurred image signals and the input image signal. You can also. As described above, when the original image signal is developed in multiple resolutions, a blurred image signal is created by interpolating and enlarging the low resolution image signal using the hierarchical image signal representing the image for each resolution as the low resolution image signal. You can also

  Here, the “blurred image signal” is an image signal that represents an image that has the same number of pixels as the input image for processing (original image signal) but has a lower sharpness (resolution) than the original image signal. It is. The blurred image signal first thins out the pixels by applying predetermined filtering processing, linear interpolation processing, or simply reducing the number of pixels to the pixels of the original image signal at predetermined intervals. The same filtering process is repeated on the obtained image signal to further reduce the number of pixels and create a plurality of low resolution image signals representing a low resolution image whose resolution is 1 / 2n times that of the original image (processing image). Each of them is created by performing an interpolation process so that the number of pixels is the same as that of the original image by a predetermined interpolation method. Here, the filtering process and the interpolation are performed by a filter having a filter coefficient determined on the basis of the resolution (for example, pixel density) of the input processing image. Various generally used methods such as the method described in Kaihei 10-63837 can be used.

  In this case, the “band-limited image signal” may be created by, for example, obtaining a difference between the blurred image signals in adjacent frequency bands, or by creating a difference between the original image signal and each blurred image signal. Good. Alternatively, the difference can be obtained by another combination of the original image signal and the blurred image signal. The creation of the converted image signal, the frequency enhancement process, and the dynamic range compression process can be expressed by, for example, the above expressions (2) and (3).

A first image processing apparatus according to the present invention is an apparatus that performs the first image processing method, that is, an image signal generation that generates a plurality of band limited image signals having different band limiting characteristics from an input image signal. A plurality of conversion functions corresponding to each of the band-limited image signals, a conversion process is performed based on the set conversion function to obtain a conversion image signal, and the input based on the conversion image signal An image processing apparatus including a conversion processing unit that performs predetermined processing on an image signal to obtain a processed image signal,
The conversion processing means holds a reference conversion function corresponding to each band-limited image signal for the reference resolution image, which is used when the conversion process is performed on the reference resolution image having the reference resolution, and has a resolution different from the reference resolution. When an image signal representing a processing image having an image is input to the image signal creating means, the frequency band of each band limited image signal corresponding to the reference conversion function is the highest in the frequency band of each band limited image signal for the processing image. The reference conversion function is relatively moved so as to be close to each other, and the relatively moved reference conversion function is set as a conversion function for the processing image.

  In the first image processing apparatus, the conversion processing means determines the number of sets of the conversion function for the processing image and each band limited image signal corresponding to the conversion function, and the resolution of the processing image is the reference resolution. When it is within the range of (p + 1) times to 2-p times (p is a positive integer), the number is reduced by p, and the resolution of the processing image is 2q-1 times to 2q times the reference resolution (q is a positive integer) ) Is preferably increased by q.

  In the first image processing apparatus, when the resolution of the processing image is lower than the reference resolution, the conversion processing means converts the conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image among the plurality of conversion functions. Can also be set as a conversion function for performing the conversion process.

  Furthermore, in the first image processing apparatus according to the present invention, in order to obtain an output image having an image size different from the image size of the reference resolution image, a reference resolution image close to the image size of the output image is further obtained. A first scaling factor of 22k (k is an integer), and a scaling factor for scaling the image scaled at the first scaling factor so that the image size of the image is equal to the image size of the output image. A magnification calculating means for calculating a magnification of 2, a first scaling processing means for obtaining a processing image by scaling the reference resolution image with the first magnification, and a processing obtained by performing the scaling process A plurality of band-limited image signals having mutually different band-limiting characteristics are created from an image signal representing an image, and a conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image is converted among the plurality of conversion functions Function and A control means for controlling the image signal generating means and the conversion processing means so as to obtain a converted image signal by performing the conversion processing by setting and obtaining the processed image signal based on the converted image signal, and the processed image It is preferable that the image processing apparatus further includes second scaling processing means for obtaining a scaled image signal representing an output image by further scaling the signal at a second magnification.

  Note that a magnification image obtained by multiplying the image size of the reference resolution image by 22k, which is closest to the image size of the output image, is obtained by the magnification calculation means and the first magnification processing means, and this magnification image is used as a processing image. Means are configured. The second scaling processing unit functions as a unit that performs scaling processing on the 22k-fold image represented by the processed image signal so as to be equal to the image size of the output image.

A second image processing apparatus according to the present invention is an apparatus that implements the second image processing method, that is, an image signal generation that generates a plurality of band limited image signals having mutually different band limiting characteristics from an input image signal. A plurality of conversion functions corresponding to each of the band-limited image signals, a conversion process is performed based on the set conversion function to obtain a conversion image signal, and the input based on the conversion image signal An image processing apparatus including a conversion processing unit that performs predetermined processing on an image signal to obtain a processed image signal,
The conversion processing means holds a reference conversion function corresponding to each band-limited image signal for the reference resolution image, which is used when the conversion process is performed on the reference resolution image having the reference resolution, and has a resolution different from the reference resolution. When an image signal representing a processing image having a processing image is input to the image signal creating means, the response of the processed image signal for the processing image is determined based on the difference between the resolution of the processing image and the reference resolution. The reference conversion function is corrected so as to be substantially the same as the response of the processed image signal for, and the function obtained by this correction is set as the conversion function for the processing image. .

  In the second image processing apparatus according to the present invention, the conversion processing means causes the image characteristic represented by the processed image signal obtained by processing the image signal representing the processing image to be the resolution of the processing image. In the frequency range of at least 1/5 or less of the corresponding Nyquist frequency, the reference conversion function is set so as to substantially match the characteristics of the image represented by the processed image signal obtained by processing the image signal representing the reference resolution image. It is desirable to perform correction.

  The first and second image processing apparatuses according to the present invention further include resolution information acquisition means for acquiring information relating to the resolution of the processing image, and the conversion processing means is provided by the resolution information acquisition means. It is desirable to set a conversion function for the processing image based on the acquired information.

  The reference conversion function in the first and second image processing apparatuses according to the present invention may be a nonlinear function.

  Further, the predetermined processing in the first and second image processing apparatuses according to the present invention may be frequency enhancement processing or dynamic range compression processing.

  The image processing method according to the present invention may be provided by being recorded on a computer-readable recording medium as a program for causing a computer to execute the image processing method.

  According to the first image processing method and apparatus of the present invention, the reference conversion function corresponding to each band limited image signal for the reference resolution image, which is used when the conversion process is performed on the reference resolution image having the reference resolution, is performed. When an image signal representing a processing image having a resolution different from the reference resolution is input, the frequency band of each band limited image signal corresponding to the reference conversion function is the frequency band of each band limited image signal for the processing image. Since the reference conversion function is relatively moved so as to be closest to the frequency band, and the moved reference conversion function is set as the conversion function for the processing image, an input image signal (reference resolution image) representing the reference resolution image is set. Signal) and an input image signal (processing image signal) representing the processing image, the same among the band limited image signals obtained from each of them Or against the band-limited image signal indicating a frequency range in the vicinity, so that enhancement processing or the like is performed in substantially the same degree, it is possible to set a conversion function for processing image. Therefore, in the output image obtained by reproducing the processed image signal obtained by performing the predetermined processing on the processing image, as long as the frequency range is substantially the same as that of the reference resolution image (image signal representing it), the frequency The response characteristics are approximately the same as the frequency response characteristics of the output image obtained by reproducing the processed image signal obtained from the reference resolution image signal. In other words, the frequency response is almost the same without being affected by the resolution of the image. A processed image signal capable of reproducing an image having characteristics can be obtained.

  Further, the number of sets of the conversion function for the processing image and each band limited image signal corresponding to the conversion function is changed from 2- (p + 1) times to 2-p times (p is a positive value). P) when the resolution is within the range of 2q-1 to 2q times the reference resolution (q is a positive integer). In addition, when the resolution of the processing image is lower than the reference resolution, a conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image among the plurality of conversion functions is set as a conversion function for performing the conversion process. Then, the same conversion function as the reference conversion function used for the reference resolution image can be used as the conversion function for the processing image, and it is not necessary to prepare a conversion function for each resolution. The configuration of the system to which the invention is applied can be simplified, and the troublesome management of the conversion function can be eliminated.

  Here, when performing frequency emphasis processing on a processing image having a resolution lower than the reference resolution, a conversion function corresponding to a frequency band equal to or lower than the resolution of the processing image is included in the reference conversion function for the reference resolution image. , Setting as a conversion function for performing the conversion process corresponds to a predetermined conversion function for each frequency band used when obtaining a processed image signal from the reference resolution image signal, corresponding to the frequency band of the processing image signal It is to perform conversion of a plurality of image signals having different band limiting characteristics obtained from the processing image signal by the converted function. When the resolution of the processing image is 1 / 2n times the reference resolution, the peak frequency of each band limited image signal obtained from the processing image signal is the peak frequency of each band limited image signal obtained from the reference resolution image signal. Therefore, the reference conversion function to be applied to the reference resolution image is obtained from the processing image signal by performing conversion processing corresponding to the frequency bands of the plurality of image signals obtained from the processing image signal. The frequency response characteristic of the converted image signal can be the same as the frequency response characteristic of the converted image signal obtained from the reference resolution image signal. For this reason, in an image obtained by reproducing the processed image signal obtained from the processing image signal, an image obtained by reproducing the processed image signal obtained from the reference resolution image signal with its frequency response characteristic. Thus, a processed image signal capable of reproducing an image having a certain frequency response characteristic that is not influenced by the resolution can be obtained.

  Furthermore, in the first image processing method and apparatus according to the present invention, when an output image having an image size different from the image size of the reference resolution image is obtained, the image size of the reference resolution image is close to the image size of the output image. When the converted image obtained by multiplying the image by 22k is converted into a processing image and subjected to the predetermined processing, the first magnification is obtained by setting the resolution of the processing image to 2k times the reference resolution. The peak frequency of each band limited image signal obtained from the image signal for use has a portion that matches the peak frequency of each band limited image signal obtained from the reference resolution image signal. Therefore, the frequency of the converted image signal obtained from the processing image signal is obtained by performing the conversion processing in such a manner that the reference conversion function applied to the reference resolution image corresponds to the frequency band of each band limited image signal obtained from the processing image signal. The response characteristic can be made substantially the same as the frequency response characteristic of the converted image signal obtained from the reference resolution image signal. For this reason, an image obtained by applying a scaling process to an image having a size of 22k times represented by the processed image signal obtained from the processing image signal so as to be equal to the desired image size of the output image. In an image obtained by reproducing a double-processed image signal, the frequency response characteristic of the image can be substantially matched with the frequency response characteristic of the image obtained by reproducing the processed image signal obtained from the reference resolution image signal. Thus, it is possible to obtain a scaled image signal that can reproduce an image having a certain frequency response characteristic that is not influenced by the resolution.

  Further, when the image size of the output image is made smaller than the image size of the reference resolution image, when the first magnification is 22k (k is a negative integer), the processing image is a pixel of the reference resolution image. The number of reduced images is smaller than the number, and the reduced image can be subjected to predetermined processing such as enhancement processing, so that the calculation time for the predetermined processing can be shortened.

  In addition, according to the second image processing method and apparatus of the present invention, the reference corresponding to each band limited image signal for the reference resolution image used when the conversion process is performed on the reference resolution image having the reference resolution. When a conversion function is prepared and an image signal representing a processing image having a resolution different from the reference resolution is input, a processed image signal for the processing image based on the difference between the resolution of the processing image and the reference resolution Since the reference conversion function is corrected so that the response is substantially the same as the response of the processed image signal for the reference resolution image, and the function obtained by this correction is set as the conversion function for the processing image, In the processed image signal obtained from the processing image signal, the frequency response characteristic is processed using the reference conversion function. In the image obtained by reproducing the processed image signal obtained by this, the frequency response characteristic of the processed image signal can be made almost the same as that of the processed image signal. It can be substantially constant. Therefore, a processed image signal capable of reproducing an image having a certain frequency response characteristic that is not influenced by the resolution can be obtained.

  In addition, since the reference conversion function is corrected, the conversion function for performing the conversion process on the processing image signal can be calculated based on the reference conversion function, and a plurality of conversion functions are prepared for each resolution. This eliminates the need to manage the conversion function.

  Further, the reference resolution image signal in the frequency range where the characteristic of the image represented by the processed image signal obtained by processing the processing image signal is at least 1/5 or less of the Nyquist frequency according to the resolution of the processing image. When the reference conversion function is corrected so that it substantially matches the image characteristic represented by the processed image signal obtained by processing the image, the frequency response characteristic in a relatively low frequency region having a large degree of influence on the image quality Can be made substantially the same as the frequency response characteristic obtained when the reference resolution image signal is processed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the present embodiment, a blurred image signal is used for an image signal obtained by reading a radiation image of the human body recorded on the stimulable phosphor sheet so that the image is suitable for diagnosis. Thus, the frequency enhancement process is performed, and the processed image signal is mainly recorded on a film and used for diagnosis.

  FIG. 1 is a schematic block diagram showing the configuration of the image processing apparatus according to the first embodiment of the present invention. The image processing apparatus 1 is a blur that generates a blurred image signal from an original image signal Sorg ′ obtained by a reading device or the like and converted by encoding the original image signal Sorg into a multi-resolution space as will be described later. The image signal generating means 2 and the frequency emphasis processing means 3 for obtaining an intermediate processed image signal Sproc ′ by performing frequency emphasis processing for emphasizing a specific frequency.

  Further, the image processing apparatus 1 includes a parameter setting unit 4, an enlargement ratio input unit 5, and an enlargement / reduction unit 6. The parameter setting means 4 is a means for setting a conversion function used by the frequency enhancement processing means 3 for the conversion process, and sets the conversion function based on the information M related to the enlargement ratio input from the enlargement ratio input means 5. The enlargement factor input means 5 is a means for obtaining the enlargement factor information M of the original image signal Sorg. The user may input the enlargement factor as a numerical value from the keyboard, and displays several kinds of enlargement factors on the operation screen and displays the user. It may be in the form of letting the user select. The enlargement / reduction means 6 enlarges / reduces the intermediate processed image signal Sproc ′ obtained by the frequency enhancement processing means 3 based on the enlargement ratio information M to obtain a final processed image signal Sproc.

  The original image signal Sorg is converted into a multi-resolution space and encoded as follows. First, as shown in FIG. 2 (a), the original image signal Sorg is wavelet transformed to be decomposed into four data LL1, HL0, LH0 and HH0 for each of a plurality of resolutions. Here, the data LL1 represents an image obtained by reducing the length and breadth of the original image to ½, and the data HL0, LH0, and HH0 represent images of vertical edges, horizontal edges, and oblique edge components, respectively. Then, the original image signal Sorg is obtained by inverse wavelet transforming the four data LL1, HL0, LH0 and HH0. Next, as shown in FIG. 2B, the data LL1 is further wavelet transformed to obtain four data LL2, HL1, LH1, and HH1. Here, the data LL2 represents an image obtained by further reducing the vertical and horizontal directions of the data LL1 to ½, and the data HL1, LH1, and HH1 represent images of vertical edges, horizontal edges, and oblique edge components of the data LL1, respectively. Become. The four data LL2, HL1, LH1, and HH1 are subjected to inverse wavelet transform to obtain a low-resolution image signal representing an image having a resolution half that of the original image. Further, the wavelet transformation is repeated as many times as desired for the data LL obtained each time the wavelet transformation is performed, thereby obtaining data for a plurality of resolutions. Thereafter, as shown in FIG. 2 (c), the data for each resolution is encoded to obtain the original image signal Sorg '.

  Only the data up to the desired resolution in the original image signal Sorg ′ is decoded and subjected to inverse wavelet transform, so that the resolution and image size are 1 / 2k (k: desired resolution) times that of the original image. The low-resolution image signal BBk representing the 1 / 22k times image can be restored. As means for performing this function, the blurred image signal creation means 2 has the same configuration as that provided with a plurality of stages of decoding means 30 (see FIG. 17) described in the second embodiment to be described later, and the restored desired state. Switching means for inputting the low resolution image signal BBk having the resolution to be input to the filtering processing means 10 is provided (both not shown in FIG. 1). Unlike the configuration provided with a plurality of decoding means 30 as shown in FIG. 17, the data LLK + 1, HLK, LHK and HHk are subjected to inverse wavelet transform and immediately represent an image having a resolution of 1 / 2k times the original image. It can also be configured to restore the low resolution image signal BBk.

  Next, the blur image signal creation process will be described in detail. FIG. 3 is a schematic block diagram showing the configuration of the blurred image signal creation means in the first embodiment.

  As shown in FIG. 3, the blurred image signal creation unit 2 in the first embodiment first performs arbitrary processing obtained by restoring the original image signal Sorg ′ as the original image signal Sorg input to the filtering processing unit 10. Using the image signal of the resolution level, the filtering processing means 10 applies a filtering process to the original image signal Sorg with respect to the x direction and the y direction of the pixel of the original image represented by the original image signal Sorg. A signal B1 is generated, and then the same filtering process is performed on the low resolution image signal B1 to generate a low resolution image signal B2 having a resolution lower than that of the low resolution image signal B1. The process is repeated. Then, the interpolation processing means 11 applies interpolation enlargement processing to the low-resolution image signal Bk obtained in each stage of the filtering processing, respectively, and a plurality of blurred image signals Sus1 to SusN (hereinafter referred to as Susk (hereinafter referred to as Susk (hereinafter referred to as “Susk”)). k = 1 to N)).

In the present embodiment, a filter substantially corresponding to a one-dimensional Gaussian distribution is used as the filter for the filtering process. That is, the filter coefficient of the filter is determined according to the following equation (4) regarding the Gaussian signal.

  This is because the Gaussian signal has good locality in both frequency space and real space. For example, a 5 × 1 one-dimensional filter in the case where σ = 1 in the above equation (4) is as shown in FIG. It will be something.

  As shown in FIG. 5, the filtering process is performed on the original image signal Sorg or on the low resolution image signal Bk every other pixel. By performing the filtering processing every other pixel in the x direction and the y direction, the number of pixels of the low resolution image signal B1 becomes 1/4 of the original image, and the low resolution image signal obtained by the filtering processing is repeated. By performing this filtering process, the n low-resolution image signals Bk (k = 1 to n) obtained become image signals whose number of pixels is 1 / 22K of the original image signal Sorg.

Next, an interpolation enlargement process performed on the low resolution image signal Bk obtained in this way will be described. Various methods such as a B-spline method can be used as the interpolation calculation method. In this embodiment, a Gaussian signal is used for the interpolation calculation. Specifically, in the following formula (5), an approximation of σ = 2k−1 is used.

  When the low resolution image signal B1 is interpolated, σ = 1 since k = 1. In this case, the filter for performing the interpolation process is a 5 × 1 one-dimensional filter as shown in FIG. This interpolation process first expands the low-resolution image signal B1 to the same size as the original image by interpolating one pixel at a time from every other pixel to the low-resolution image signal B1. This is performed by subjecting the interpolated low-resolution image signal B1 to a filtering process using the one-dimensional filter shown in FIG.

  Similarly, this interpolation enlargement process is performed for all the low resolution image signals Bk. When interpolating the low-resolution image signal Bk, a filter having a length of 3 × 2k −1 is created based on the above equation (5), and pixels having a value of 0 are inserted between the pixels of the image signal Bk. By interpolating 2k −1 at a time, the image is enlarged to the same size as the original image, and filtering processing is performed by a filter having a length of 3 × 2k −1 on the image signal Bk interpolated with a pixel having this value of 0 The interpolation is enlarged by applying.

  Next, frequency enhancement processing performed using the blurred image signal Susk created as described above will be described. FIG. 7 is a schematic block diagram showing the configuration of the apparatus that performs the frequency enhancement process together with the blurred image signal creating means 2. As shown in FIG. 7, the difference between the original image signal Sorg and the blurred image signal Susk created by the filtering processing means 10 and the interpolation processing means 11 is obtained by the subtractor 21, and the original image signal Sorg Band-limited image signals (Sorg-Sus1, Sus1-Sus2, etc.), which are limited frequency band components, are created.

  On the other hand, the parameter setting means 4 provided in the image processing apparatus 1 uses reference conversion corresponding to each band limited image signal for the reference resolution image, which is used when conversion processing is performed for the reference resolution image having the reference resolution. Functions f1 to fN are held in a memory (not shown). When the reference resolution image signal is input as the original image signal Sorg input to the filtering processing means 10, the held reference conversion functions f1 to fN are set according to each band limited image signal.

When the reference resolution image signal is input to the filtering processing means 10, each band limited image signal is converted by the converter 22 so as to have a desired magnitude by different conversion functions f1 to fN. 2 '), the plurality of converted band-limited image signals (converted image signals of the present invention) are integrated in the arithmetic unit 23 to obtain a signal (integrated signal) Fusm necessary for frequency enhancement processing, and further the signal Fusm is multiplied by a predetermined coefficient and then added to the original image signal Sorg to generate a processed image signal Sproc '. As a result, it is possible to obtain a processed image signal Sproc ′ in which a desired frequency component is enhanced to a degree according to the purpose.
Sproc ′ = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2,... SusN)
Fusm (Sorg, Sus1, Sus2, ... SusN)
= F1 (Sorg-Sus1) + f2 (Sus1-Sus2) +
+ Fk (Susk-1-Susk) + ... + fN (SusN-1-SusN) (2 ')
(However, Sproc ': Image signal with enhanced high-frequency components
Sorg: Original image signal
Susk (k = 1 to N): Blur image signal
fk (k = 1 to N): conversion function for converting each band limited image signal
β (Sorg): enhancement coefficient determined based on the original image signal

  The creation of the processed image signal Sproc ′ has been described above. Next, the problem to be solved by the first embodiment and the solution means will be described with an example. As described above, when the processed image signal Sproc ′ representing an image having the same resolution as the original image (which may be the same image as the original image or a different image) is obtained, the frequency of the band limited image signal The frequency enhancement process may be performed using the conversion function fk according to the band, but the processed image representing an image with a lower resolution than the original image (which may be the same image as the original image or a different image) When an image signal is obtained, it is necessary to prepare a conversion function according to the resolution, so that management of the conversion function becomes very troublesome. Here, when the frequency enhancement process is performed by separating the original image signal Sorg into six frequency bands, the frequency characteristics of the band limited image signal have six peaks.

  For example, an original image signal Sorg representing a reference resolution image read at a reading density of 10 lines / mm as a reference resolution (hereinafter also referred to as an original image signal having a resolution of 10 lines / mm; also at other reading densities) In the case of obtaining a band-limited image signal from the same), the Nyquist frequency of the original image signal Sorg is 5 cycles / mm, and the peak frequency of the band-limited image signal in the highest frequency band is 5 cycles / mm as shown in FIG. The peak frequency of the band limited image signal in the frequency band next to the highest frequency band is 1.0 cycle / mm having a value that is 1/5 of the Nyquist frequency, and is 0.5 cycle / mm as the frequency band becomes lower. The peak frequency is ½ of the peak frequency of the previous stage, such as 0.25 cycle / mm, 0.12 cycle / mm, and 0.06 cycle / mm. Also, an image signal representing an image having a resolution half that of the reference resolution image (the reading density may be 5 lines / mm or the reference resolution image may be reduced by thinning or the like) is referred to as an original image signal Sorg. The peak frequency of the band-limited image signal is 2.5 cycle / mm, 0.5 cycle / mm, 0.25 cycle / mm, 0.125 cycle / mm, and 0.06 cycle / mm as shown in FIG. .

  The peak frequency of the band limited image signal when the image signal representing the 1/4, 1/8, 1/16 resolution image of the reference resolution image is the original image signal Sorg is shown in FIGS. It becomes like this. That is, the peak frequency of the band limited image signal in the highest frequency band among the band limited image signals is the Nyquist frequency corresponding to the resolution level, and the peak frequency of the band limited image signal in the frequency band next to the highest frequency band is 1 of the Nyquist frequency. / 5, and becomes a half of the peak frequency in the previous stage as the frequency band becomes lower.

  Next, a case where an output image having an image size different from the image size of the reference resolution image (hereinafter also referred to as a desired image size) is obtained will be described.

  For example, when a processed image signal representing an image having a desired image size smaller than the reference resolution image is obtained, an image signal representing a processing image according to the present invention, that is, an original image signal Sorg input to the filtering processing means 10 is used. An image signal representing a low-resolution image that is a scaled image obtained by multiplying the image size of the reference resolution image closest to the desired image size by 22k (k is a negative integer) is used. Hereinafter, the resolution of a low resolution image obtained by multiplying the image size by 22k is referred to as a predetermined resolution, and an image signal representing the low resolution image is also referred to as a reference low resolution image signal.

  Then, a low-resolution blurred image signal is obtained from the reference low-resolution image signal, a low-resolution band-limited image signal is created based on the low-resolution blurred image signal, and a band-limited image signal with respect to the low-resolution band-limited image signal A converted image signal is obtained by a conversion function corresponding to the frequency band of the image, an intermediate processed image signal Sproc 'is obtained, and the intermediate processed image signal Sproc' is enlarged or reduced based on the enlargement ratio information M to obtain a desired image size. A processed image signal Sproc representing the image is obtained.

  As a more specific mode, a case where a processed image signal Sproc representing an image of ½ resolution (1/22 reduced image) with an image size ¼ of the reference resolution image is obtained will be described with reference to the flowchart of FIG. I will explain. Here, in the original image signal Sorg, the band-limited image signal is assumed to be composed of six frequency bands as shown in FIG.

  First, an enlargement factor (desired enlargement factor) corresponding to the desired image size desired by the user is inputted from the enlargement factor input means 5 (step S1). As a result, the enlargement ratio information M is input from the enlargement ratio input means 5 to the blurred image signal creation means 2, the parameter setting means 4, and the enlargement / reduction means 6. Based on the enlargement ratio information M, the blurred image signal creation means 2 decodes a low resolution image signal BBk representing an image whose image size closest to the desired enlargement ratio is 22k (k is a negative integer) times. 17) and is input to the filtering processing means 10 (step S2), and a low resolution blurred image signal Susk is created (step S3). The low resolution image signal BBk becomes a reference low resolution image signal corresponding to the original image signal Sorg. In the present embodiment, since enlargement ratio information M corresponding to an image size of 1/4 of the reference resolution image is input, the low-resolution image signal BB1 representing an image having an image size of 1/4 times. To restore.

  Here, six blur image signals (including the original image signal) Sorg to Sus6 obtained when the reference resolution image signal is input to the filtering processing means 10 as the original image signal Sorg, which is close to a predetermined resolution (desired image size). The correspondence relationship between the low-resolution blurred image signals Sorg to Sus5 obtained from the low-resolution image signal BB1 representing the 1/2 resolution image (1/22 reduced image) is the same in the same frequency range of each blurred image signal. As shown in FIG. FIG. 14 also shows the correspondence between the original image signal Sorg and the blurred image signal Susk from a 1/4 resolution image (1/42 reduced image) to a 1/16 resolution image (1/162 reduced image). As shown in FIG. 14, the blurred image signal Sus1 in the reference resolution image corresponds to, for example, the low resolution blurred image signal Sorg in the ½ resolution image, and sequentially moves relative to the right side in the figure.

  Next, a low resolution band limited image signal is created based on the low resolution blurred image signal Susk (step S4). Correspondence relationship between the six band limited image signals obtained from the reference resolution image signal and the low resolution band limited image signal obtained from the low resolution image signal BBk representing each resolution image is the same in the frequency range of each band limited image signal. It shows in FIG. 15 so that things may become the same row. Of the low-resolution band limited image signals generated when the low-resolution image signal BBk is input to the filtering processing means 10, those having a low frequency band of 0.03 cycle / mm or less are emphasized for the reference resolution image. Since it is a signal in a frequency range that is not used for processing, it is not used for enhancement processing. This is why the right side of the 1/2 to 1/16 resolution images in FIGS.

  On the other hand, the parameter setting means 4 sets the conversion function fk for the conversion process performed on each low resolution band limited image signal (step S5). Here, as described above, the conversion functions f1 to f6 are used for the six band limited image signals obtained from the reference resolution image signal, but the low resolution image signal BB1 representing the 1/2 resolution image is used. For a resolution band limited image signal, a reference is made so that the peak wavelength of the low resolution band limited image signal corresponding to the conversion function to be set is the same as the wavelength of each band limited image signal obtained from the reference resolution image signal. The conversion functions f1 to f6 are shifted (relative movement).

FIG. 16 shows the correspondence between the conversion functions in each resolution image. For example, as can be seen from FIG. 8, FIG. 9, and FIG. 14, for example, the peak wavelength of the band limited image signal Sus1-Sus2 in the reference resolution image is the peak of the low resolution band limited image signal Sorg-Sus1 in the 1/2 resolution image. It becomes the same as the wavelength, and subsequently moves relative to the right side in FIG. Therefore, as shown in FIG. 16, when the resolution of the low resolution image as the processing image is 1 / 2k times the reference resolution, the conversion functions fN for the low resolution image are k pieces of the reference conversion functions f1 to f6. The one on the low frequency band side is used. Therefore, when the integration signal Fusm in the above equation (2 ′) when the frequency enhancement process is performed on the reference resolution image is shown in the expression (6), the integration when the frequency enhancement process is performed on the ½ resolution image. The signal Fusm is as shown in equation (7).
Fusm (Sorg, Sus1, Sus2, ... Sus6)
= F1 (Sorg-Sus1) + f2 (Sus1-Sus2) +
+ F5 (Sus4−Sus5) + f6 (Sus5−Sus6) (6)
Fusm (Sorg, Sus1, Sus2, ... Sus5)
= F2 (Sorg-Sus1) + f3 (Sus1-Sus2) + ...
+ F4 (Sus3-Sus4) + f5 (Sus4-Sus5) (7)

  Then, when the low resolution converted image signal is generated from the low resolution band limited image signal in this way (step S6) and the low resolution integrated signal Fusm is obtained, the enhancement coefficient β as shown in the above equation (2 ′). (Sorg) is multiplied by the low-resolution integrated signal Fusm and further added to the original image signal Sorg (here, the reference low-resolution image signal) to create an intermediate processed image signal Sproc '(step S7). This intermediate processed image signal Sproc 'represents an image of 1/2 resolution and 1/2 resolution image (1/22 reduced image) of the reference resolution image. Note that β (Sorg) in equation (2 ′) is set based on the reference low-resolution image signal.

  The intermediate processed image signal Sproc ′ obtained in this way is input to the enlargement / reduction means 6. The enlargement / reduction means 6 finally enlarges or reduces the intermediate processed image signal Sproc ′ based on the enlargement ratio information M input from the enlargement ratio input means 5 so that an image with a desired enlargement ratio can be reproduced. A processed image signal Sproc is obtained (step S8). In this embodiment, since the enlargement ratio information M input from the enlargement ratio input unit 5 corresponds to an image size of 1/4, the enlargement / reduction unit 6 uses a 1/22 reduced image (1/2 resolution). The intermediate processed image signal Sproc ′ representing the image) is used as the final processed image signal Sproc without any enlargement or reduction.

  Thus, in the present embodiment, when obtaining a processed image signal representing an image having an image size smaller than the reference resolution image, or obtaining a processed image signal representing an image having a lower resolution than the reference resolution image, The reference conversion function used when performing frequency enhancement processing on the reference resolution image is the frequency band of each band limited image signal for the reference resolution image closest to the frequency band of each band limited image signal for the processing image. The reference conversion function is relatively moved so that the frequency conversion processing is performed as a result of obtaining the converted image signal by setting the reference conversion function thus moved as a conversion function for the processing image. In the image obtained by reproducing the processed image signal, the frequency response characteristic is processed image signal Sproc based on the reference resolution image. The processed image signal Sproc that can reproduce an image having a certain frequency response characteristic that is not influenced by the resolution can be obtained.

  In addition, since the same conversion function as that used when performing frequency emphasis processing on the reference resolution image is used, it is not necessary to prepare a conversion function for each resolution, and thus the system to which the present invention is applied. Can be simplified, and the troublesome management of the conversion function can be eliminated.

  Even when a processed image signal representing an image having a resolution lower than that of the reference resolution image is obtained regardless of the number of pixels (image size), the reference conversion functions f1 to f6 may be similarly shifted, and the obtained effect. Is almost the same. However, since the sharpness cannot be improved by the processing in the enlargement / reduction means 6, the final image may not be completely restored to the desired resolution although the image size becomes the desired image size.

  Further, when the image size of the output image is made smaller than the image size of the reference resolution image, when the size of the image input to the filtering processing means 10 is 22k (k is a negative integer) times the reference resolution image. The processing image becomes a reduced image having a number smaller than the number of pixels of the reference resolution image, and the reduced image can be subjected to predetermined processing such as enhancement processing, so that the calculation time for the predetermined processing can be shortened. it can.

  In the first embodiment, a blurred image signal is obtained from the reference resolution image signal or the reference low resolution image signal by filtering and interpolation expansion, and a band limited image signal is created from the reference resolution image signal and the blurred image signal. However, the present invention is not limited to this. For example, a reference resolution image signal or a reference low resolution image signal is converted into multiple resolutions by a method such as a wavelet transform or a Laplacian pyramid, and the converted image signal for each resolution is converted into a band. It may be a restricted image signal.

  In the first embodiment, the frequency enhancement process is performed after the image of the predetermined resolution is restored based on the original image signal Sorg ′ converted into the multi-resolution and encoded. However, the present invention is not limited to this. However, frequency enhancement processing may be performed on an image signal that is not encoded at all or an image signal having a resolution different from that of the image signal that is not encoded.

  Next, a second embodiment of the present invention will be described. In the first embodiment, the original image signal Sorg ′ obtained by converting the original image signal Sorg into a multi-resolution space and encoding the original image signal Sorg ′ that is closest to the desired image size is a 1 / 2k resolution image (1 / 22k reduced image). In the second embodiment, the low-resolution image signal BBk representing a plurality of blurred image signals is generated by sequentially repeating filtering and interpolation expansion on the restored low-resolution image signal BBk. A plurality of low resolution image signals BBk of each resolution level (each image size) from the resolution (minimum image size) to the image of 1 / 2k resolution (1 / 22k reduced image) closest to the desired image size are interpolated and enlarged. This creates a blurred image signal. In the second embodiment, only the process in the blurred image signal creating unit 2 is different, and the other configuration is the same as that of the first embodiment, and thus detailed description thereof is omitted.

  Here, a blur image signal creation process in the second embodiment will be described. FIG. 17 is a schematic block diagram showing the configuration of the blurred image signal creation means 2 in the second embodiment. As shown in FIG. 17, first, the data representing the lowest resolution image in the original image signal Sorg ′ is decoded and inverse wavelet transformed by the decoding means 30 to obtain a low resolution representing an image having a resolution of 1 / 2n times the original image. An image signal BBn is generated, and then a low resolution image signal BBn-1 representing an image having a resolution 1 / 2n-1 times that of the original image is generated using the low resolution image signal BBn. Wavelet transform is performed to generate a low resolution image signal BBk that represents a resolution 1 / 2k (k = 1 to n) times the original image and an original image signal Sorg that represents the highest resolution image, that is, the original image (reference resolution image). . The low resolution image signal BBk corresponds to data LLk obtained by wavelet transforming the original image signal Sorg. Then, the interpolation processing means 31 performs an interpolation enlargement process similar to that in the first embodiment on the low resolution image signal BBk obtained at each stage of the decoding process, and a plurality of different sharpness levels are obtained. Blurred image signals Sus1 to SusN (hereinafter represented by Susk (k = 1 to N)) are obtained.

  Next, a band limited image signal is obtained from the blurred image signal Susk. Then, as shown in the above equation (2 ′), a frequency enhancement process is performed to obtain a processed image signal Sproc ′.

  Here, also in the second embodiment, a processed image representing an image having a desired image size smaller than the reference resolution image (same meaning as an image having a predetermined resolution lower than the reference resolution), as in the first embodiment. A case where the signal Sproc is obtained will be described. In the second embodiment, it is assumed that the frequency band of the band limited image signal generated in the same manner as in the first embodiment is six, and the desired image size is a quarter size of the reference resolution image.

  First, the user inputs a desired enlargement ratio from the enlargement ratio input means 5. As a result, the enlargement ratio information M is input from the enlargement ratio input means 5 to the blurred image signal creation means 2, the parameter setting means 4, and the enlargement / reduction means 6. Based on the enlargement ratio information M, the blurred image signal creation unit 2 generates images of each image size (each resolution) from the lowest resolution up to 22k (k is a negative integer) times the image size closest to the desired enlargement ratio. The low resolution image signal BBk to be represented is restored and the low resolution blurred image signal Susk is created. The low resolution image signal BBk becomes a reference low resolution image signal corresponding to the original image signal Sorg. In the present embodiment, since enlargement ratio information M corresponding to an image size of 1/4 of the reference resolution image is input, the low-resolution image signal BBn representing an image having an image size of 1/4 times. Restore ~ BB1.

  Here, the generated six blurred image signals (including the original image signal) Sorg to Sus6 and the low resolution image representing the 1/2 resolution image (1/22 reduced image) closest to the predetermined resolution (desired image size). Correspondence relationship between the low-resolution blurred image signals Sorg to Sus5 obtained from the signal BB1 and the relationship between the band-limited image signal obtained from the reference resolution image signal and the band-limited image signal obtained from the low-resolution image signal BBk representing each resolution image Is as shown in FIG. 14 and FIG. In the parameter setting means 4, the setting of the conversion function of the conversion process performed on each low resolution band limited image signal is performed in the same manner as in the first embodiment, and when the resolution is 1 / 2k, the conversion function fN As shown in the above equations (6) and (7), k of the reference conversion functions f1 to f6 are used on the low frequency band side.

  When the low-resolution converted image signal is generated from the low-resolution band-limited image signal in this way and the low-resolution integrated signal is obtained, the enhancement signal β (Sorg) is multiplied by the integrated signal, and an image with a predetermined resolution is further obtained. An intermediate processed image signal Sproc 'is obtained by adding to the corresponding original image signal Sorg (reference low resolution image signal).

  The intermediate processed image signal Sproc ′ obtained in this way is input to the enlargement / reduction means 6. The enlargement / reduction means 6 finally enlarges or reduces the intermediate processed image signal Sproc ′ based on the enlargement ratio information M input from the enlargement ratio input means 5 so that an image with a desired enlargement ratio can be reproduced. A processed image signal Sproc is obtained. Here, also in the second embodiment, the enlargement factor information M input from the enlargement factor input unit 5 corresponds to an image size of ¼. Therefore, the enlargement / reduction unit 6 uses a 1/22 reduced image (1 / The intermediate processed image signal Sproc ′ representing the (2 resolution image) is used as the final processed image signal Sproc without any enlargement or reduction.

  As described above, also in the second embodiment, when obtaining a processed image signal representing an image having an image size smaller than the reference resolution image, the reference conversion used when the frequency enhancement process is performed on the reference resolution image. Since the band-limited image signal is converted by shifting (relatively moving) the function according to the desired image size (predetermined resolution), the frequency of the image obtained by reproducing the obtained processed image signal is The response characteristic can be made to substantially match the frequency response characteristic of the image obtained by reproducing the processed image signal Sproc based on the reference resolution image, thereby reproducing an image having a constant frequency response characteristic that is not influenced by the resolution. A processed image signal Sproc can be obtained.

  In the first and second embodiments, a case where a 1/2 reduced image is obtained has been described. For example, when a 1/32 reduced image (1/3 resolution image) is obtained, a blurred image is obtained. In the signal generating means 2 and the frequency enhancement processing means 3, frequency enhancement processing is performed based on a low resolution image signal representing a 1/42 reduced image (1/4 resolution image) closest to the 1/32 reduced image, and intermediate processing is performed. A finished image signal Sproc 'is obtained. Then, the intermediate processed image signal Sproc ′ is enlarged by (4/3) 2 times in the enlargement / reduction means 6 to obtain the final processed image signal Sproc. If the desired image size is 1 / 22k times the reference resolution image, the enlargement / reduction means 6 can be omitted, and the configuration of the apparatus can be simplified.

In the first and second embodiments, the converted image signal is created by converting the band limited image signal by the conversion function fk based on the equation (2 ′). As shown in FIG. 5, the band limited image signal may be emphasized using the enhancement coefficient αk instead of the conversion function fk to obtain a converted image signal. In this case, the emphasis processing is performed by shifting the emphasis coefficients α1 to α6 set for the reference resolution image according to the desired image size (predetermined resolution) in the same manner as the conversion function fk is shifted. .
Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2,... SusN)
Fusm (Sorg, Sus1, Sus2, ... SusN)
= Α1 (Sorg−Sus1) + α2 (Sus1−Sus2) +
+ Αk (Susk-1-Susk) + ... + αN (SusN-1-SusN) (8)

  Each of the above embodiments is for the case where the desired image size (predetermined resolution) is smaller than the reference resolution image. A method substantially similar to that of the embodiment can be used. Hereinafter, this case will be described.

  In medical diagnosis or the like, when an image is viewed on a CRT and a part of the image is interpreted in detail, the corresponding part of the image may be enlarged and displayed or output to a film. In this case, in the conventional mode, the enlarged image signal is processed using the conversion function for the reference resolution image to perform the enhancement process. On the other hand, in the enlargement process, an interpolation process is performed to obtain a 22n-fold enlarged image (n is a positive integer), and a 2n-fold resolution level signal prepared in advance for the enlarged image is used as the multi-resolution signal. You could think so. Here, the frequency band of the image signal representing the enlarged image is the same as the frequency band of the reference resolution image at the maximum although the number of pixels is 22n times in the former case, and the frequency band of the 2n times resolution level in the latter case. Expands as much as the signal is used.

  On the other hand, the generation process, the conversion process, and the enhancement process of the band limited image signal are the same regardless of the mode of the input image signal, and the interpolation process is performed in the enlargement process or the 2n times resolution level is applied. Regardless of whether a signal is used, the number of sets of band-limited image signals and conversion functions to be processed increases between the image size and the band-limited image signal as estimated from FIG. It is necessary to define a conversion function that does not exist.

  On the other hand, the frequency component carried by each band-limited image signal in the enlarged image has the same frequency range in the same column portion shown in FIG. 18 when a signal of 2n times resolution level is used in the enlargement process. For example, Sorg-Sus1 in the reference resolution image is the same as Sus1-Sus2 in the 2 times resolution image (22 times enlarged image) and Sus2 -Sus3 in the 4 times resolution image (42 times enlarged image). Then, Sorg-Sus1 in the double-resolution image and Sorg-Sus1 and Sus1-Sus2 in the quadruple-resolution image are high-frequency components (high-frequency band limited image signals) that do not exist in the reference resolution image. Therefore, when performing enhancement processing using a 2n-fold resolution level signal prepared for an enlarged image, as shown in FIG. 19, for the same frequency component as that of the reference resolution image, that for the reference resolution image is used. The same conversion function is used, and the conversion function f1 used in the highest frequency band of the reference resolution image is used as it is as a conversion function corresponding to a high frequency component that does not exist in the reference resolution image. As a result, the response of the processed image signal Sproc carrying the enlarged image is made substantially the same as the reference resolution image from the low frequency side having a large visual influence to the highest frequency (Nyquist frequency) of the reference resolution image. In addition, high-frequency components that do not exist in the reference resolution image can be subjected to enhancement processing having a certain degree of correspondence with the reference resolution image by using the conversion function f1.

  On the other hand, when interpolation processing is performed to obtain a 22n-fold enlarged image and enhancement processing is performed using an image signal representing this enlarged image, the peak wavelength of each band limited image signal in the enlarged image is, for example, that in the reference resolution image Sorg-Sus1 is the same as Sus1-Sus2 in the 22-fold magnified image and Sus2-Sus3 in the 42-fold magnified image. Then, Sorg-Sus1 in the 22-fold magnified image and Sorg-Sus1 and Sus1-Sus2 in the 42-fold magnified image are the same as the highest frequency (Nyquist frequency) of the reference resolution image. Therefore, when performing enhancement processing using the 22n-fold enlarged image obtained by performing the interpolation processing, the same conversion function as the reference resolution image is used for the same frequency component as the reference resolution image. The same conversion function as 19 is used. That is, the conversion function f1 is used for Sus1-Sus2 in the 22-fold magnified image, Sus2-Sus3 in the 42-fold magnified image, Sorg-Sus1 in the 22-fold magnified image, Sorg-Sus1, Sus1-Sus2 in the 42-fold magnified image. . As a result, the response of the processed image signal Sproc carrying the enlarged image is changed from the low frequency side having a large visual influence to the highest frequency of the reference resolution image (in this example, substantially the entire band). It can be almost the same as the image.

  Next, a third embodiment of the present invention will be described. FIG. 20 is a schematic block diagram showing the configuration of the image processing apparatus according to the third embodiment of the present invention. In FIG. 20, elements that are the same as those of the first embodiment are given the same reference numerals, and descriptions thereof are omitted unless particularly required.

  The image processing apparatus 1 according to the third embodiment includes a blurred image signal generating unit 2 that generates a blurred image signal from an original image signal Sorg having an arbitrary resolution obtained by a reading device and the above-described equation (2). Conversion processing means 3 having the same function as the frequency enhancement processing means 3 according to the first embodiment, which obtains a processed image signal Sproc by performing frequency enhancement processing for emphasizing a specific frequency based on. Furthermore, the image processing apparatus 1 according to the third embodiment includes a conversion function calculation unit 40 and a resolution information input unit 50. The conversion function calculation means 40 is a means for calculating a conversion function used by the conversion processing means 3 for the conversion process, and a reference having a reference resolution based on resolution information input from the resolution information input means 50 as will be described later. A conversion function corresponding to the resolution of the image represented by the input image signal is calculated using a conversion function used when frequency enhancement processing is performed on the resolution image signal (hereinafter referred to as a reference original image signal) Sorg0. It is. The resolution information input means 50 is means for obtaining resolution information M of the original image signal Sorg. The input by the resolution information input means 50 may be input by the user as a numerical value from the keyboard, or may be in a form in which several types of density are displayed on the operation screen and selected by the user. Alternatively, the reading apparatus may attach the resolution information M to the original image signal Sorg, and the image processing apparatus 1 may recognize the attached resolution information M for each input image signal. Any form may be used as long as the means 40 can recognize the resolution.

Note that the conversion functions f1 to fN used in the converter 22 of the third embodiment are constants. Here, as shown in FIG. 8 of the first embodiment, the conversion function set for each band limited image signal obtained from the original image signal Sorg read at a reading density of 10 lines / mm is as follows. It shall have a value as shown in Table 1.

  Next, the problem to be solved by the third embodiment and the means for solving the problem will be described with an example. As described above, in order to enhance the desired frequency component of the original image signal Sorg, the frequency enhancement process may be performed using the conversion function fk corresponding to the frequency band of the band limited image signal. However, when the conversion function used when performing the frequency emphasis processing on the original image signal Sorg having a resolution of 10 lines / mm is applied to the original image signal Sorg having a resolution of 300 dpi, 10 lines / mm. FIG. 21 shows the frequency response characteristics of the image obtained in the processed image signal Sproc obtained from the original image signal Sorg having a resolution of λ and the processed image signal Sproc obtained from the original image signal Sorg having a resolution of 300 dpi. Will be different.

  Therefore, in the third embodiment, for example, an original image signal Sorg having a resolution of 10 lines / mm is set as a reference original image signal Sorg0, and a reference conversion used when frequency enhancement processing is performed on the reference original image signal Sorg0. Using this function, a conversion function used when frequency enhancement processing is performed on an original image signal Sorg of 300 dpi is calculated. Hereinafter, calculation of the conversion function will be described.

  FIG. 22 is a diagram showing the frequency response characteristics of the band-limited image signal obtained from the 300 dpi resolution original image signal Sorg. As shown in FIG. 22, the Nyquist frequency of the original image signal Sorg with 300 dpi resolution is 5.9 cycle / mm, and the peak frequency of the band limited image signal in the highest frequency band is this Nyquist frequency. The peak frequency of the band-limited image signal in the frequency band next to the highest frequency band is 1.18 cycle / mm, which is 1/5 of the Nyquist frequency, and 0.59 cycle / mm as the frequency band becomes lower. The peak frequency has a half of the previous frequency band, such as 0.30 cycle / mm, 0.15 cycle / mm, and 0.07 cycle / mm. This relationship coincides with the relationship of the frequency response characteristics of the band limited image signal of the reference original image signal Sorg0 shown in FIG. 7, and this relationship is the same as long as the filter for creating the low resolution image signal is not changed. This is true regardless of the resolution of the image signal Sorg.

Therefore, based on the peak frequency of the band-limited image signal obtained from the reference original image signal Sorg0, the peak frequency of the band-limited image signal obtained from the original image signal Sorg having a 300 dpi resolution, and the reference conversion function, linear interpolation or extrapolation is performed. By performing the interpolation, a conversion function for a 300 dpi image can be obtained as shown in Table 2 below. Specifically, the conversion function corresponding to the band limited image signal with the peak frequency of 1.18 cycle / mm is the peak frequency and the conversion function value from the conversion function values of 5 cycle / mm and 1 cycle / mm in Table 1. And the value of the conversion function at the peak frequency is obtained by substituting the value of the peak frequency of 300 dpi resolution into the equation (9). Then, this linear interpolation is performed for all peak frequencies to obtain a conversion function corresponding to each band limited image signal. In addition, although calculated | required by linear interpolation here, it is not limited to this.

fx = 1/40 × C + 0.875 (9)
Where fx: value of the conversion function to be obtained
C: Peak frequency

  Then, the conversion processing means 3 performs conversion processing (frequency emphasis processing) by the conversion function thus obtained, and a processed image signal Sproc is obtained. Here, the processed image signal Sproc0 obtained by processing the processed image signal Sproc processed by the conversion function obtained as described above and the reference original image signal Sorg0 using the reference conversion function. FIG. 23 shows the relationship between the frequency response characteristics of and. As shown in FIG. 23, according to the first embodiment, a processed image signal Sproc having substantially the same frequency response characteristics can be obtained regardless of the resolution of the original image signal to be processed.

  Next, the operation of the third embodiment will be described. FIG. 24 is a flowchart showing the operation of this embodiment. First, an original image signal Sorg is input to the image processing apparatus 1 from a reading device or the like (step S1). The original image signal Sorg is input to the blurred image signal creating means 2 where the blurred image signal Susk is created (step S2). On the other hand, resolution information M of the original image signal Sorg is input to the resolution information input means 50 (step S3). Based on the input resolution information M, the conversion function calculation means 40 performs conversion according to the resolution as described above. A function is calculated (step S4). In addition, although the process of step S3, 4 may be performed previously rather than the process of step S1, 2, calculation time can be shortened by performing these processes in parallel. In the conversion processing unit 3, a band limited image signal is created based on the blurred image signal Susk, and further, the frequency enhancement processing shown in the above equation (2) is performed based on the conversion function calculated in the conversion function calculation unit 40. A processed image signal Sproc is obtained (step S5).

  This frequency enhancement process creates an addition signal to be added to the original image signal for enhancement using the blurred image signal, and this addition is performed in order to prevent artifacts from occurring by performing the frequency enhancement process. A different conversion process is performed for each frequency band so that each frequency band signal constituting the signal becomes a desired signal. In order to create a desired signal, it is preferable that the signal has a desired frequency response characteristic regardless of the resolution of the original image signal Sorg. According to the third embodiment, since the conversion function is calculated so that the frequency response characteristics of the processed image signal Sproc are substantially the same regardless of the resolution, the conversion function is substantially the same regardless of the resolution of the original image. A processed image signal Sproc having frequency response characteristics can be obtained.

  Moreover, since the conversion function is calculated for each resolution based on the reference conversion function, it is not necessary to prepare a plurality of conversion functions for each resolution, thereby eliminating the troublesome management of the conversion function.

  In the third embodiment, the conversion function corresponding to an arbitrary resolution is calculated using only one set corresponding to the resolution of 10 lines / mm as the reference conversion function. A conversion function corresponding to the reference resolution is prepared as a reference conversion function, and it is determined whether the resolution of the image to be processed has a value close to any of the above-mentioned several kinds of reference resolutions. A conversion function corresponding to the resolution of the image to be processed may be obtained. For example, two types of reference resolutions and conversion functions corresponding to these are used as first and second reference resolutions, first and second reference conversion functions, and the usage ranges of the first and second reference conversion functions are illustrated. In the case of setting as shown in FIG. 25 (a), when the resolution of the image represented by the original image signal to be processed is a portion indicated by an arrow A in the figure, the first reference conversion function is used, and the arrow B In the case of the second part, the second reference conversion function is used. In addition, when the usage range of the first and second reference conversion functions is set as shown in FIG. 25B, the resolution of the image represented by the original image signal to be processed is indicated by an arrow C in the figure. In some cases, the first reference conversion function may be used, and in the case of the part of the arrow D, the second reference conversion function may be used.

In the third embodiment, the conversion function is a constant, but a nonlinear function having a predetermined slope may be used. Hereinafter, calculation of the conversion function according to the resolution when the conversion function is a nonlinear function will be described as a fourth embodiment. In the fourth embodiment, a nonlinear function represented by the following equation (10) is used as the conversion function fk.
f (Sin) = Sout = Sin × Y × (exp (X / Sin) −1) / (exp (X / Sin) +1) (10)

In Expression (10), Sin is an input signal, Sout is an output signal, X is a parameter that determines the degree of nonlinearity, that is, a suppression condition, and Y is a parameter that controls the slope of the entire function, that is, the frequency response characteristic. Then, by adjusting the parameters X and Y, the frequency response characteristics of the band limited image signal can be changed. In the fourth embodiment, parameters X and Y shown in Table 3 below are used. In Table 3, parameters for original image signals having pixel densities of 10 lines / mm, 6.7 lines / mm, and 5 lines / mm are shown. Parameters in the upper row are parameters for band-limited image signals in a high frequency band. It has become. Here, the lower the Nyquist frequency, the smaller the number of band-limited image signals used for nonlinear processing as shown in FIG. For example, six band-limited image signals are used for a resolution of 10 lines / mm, but only five band-limited image signals are used for a resolution of 6.7 lines / mm and 5 lines / mm. This is because there is no low frequency band corresponding to 10 lines / mm for the low frequency band having a peak at 0.03 cycle / mm at pixel densities of 6.7 lines / mm and 5 lines / mm. In this case as well, the resolution of the reference original image signal is 10 lines / mm, and a conversion function composed of a set of parameters X and Y shown in Table 3 is used as the reference conversion function. Further, the resolution of the image represented by the original image signal to be processed is set to 200 dpi.

  FIG. 27 is a diagram schematically showing the frequency response characteristics of the band limited image signal obtained from the 200 dpi resolution original image signal Sorg. As shown in FIG. 27, the Nyquist frequency of the 200 dpi resolution original image signal Sorg is about 4.0 (strictly 3.937) cycle / mm, and the peak frequency of the band limited image signal in the highest frequency band is the Nyquist frequency. It becomes frequency. And the peak frequency of the band limited image signal in the frequency band next to the highest frequency band is 0.8 cycle / mm, which is 1/5 of the Nyquist frequency. The peak frequency has a half of the frequency band of the previous stage, such as 0.2 cycle / mm, 0.1 cycle / mm, and 0.05 cycle / mm. This relationship coincides with the relationship of the frequency response characteristics of the band limited image signal of the reference original image signal Sorg0 having a resolution of 10 lines / mm shown in FIG. 8, and this relationship creates a low resolution image signal. This is true regardless of the resolution of the original image signal Sorg unless the filter at that time is changed. The processed image signal Sproc obtained by processing the original image signal Sorg representing the 200 dpi image having such frequency response characteristics by the reference conversion function and the reference original image signal Sorg0 are processed by the reference conversion function. FIG. 28 shows frequency response characteristics with the processed image signal Sproc0 obtained by applying the frequency response. As shown in FIG. 28, the processed image signal Sproc is an image in which the low frequency band is emphasized as compared with the processed image signal Sproc0.

Therefore, as in the third embodiment, a conversion function used for processing is calculated from the reference conversion function according to the resolution of the original image signal to be processed. Hereinafter, calculation of the conversion function will be described. In the fourth embodiment, the peak frequency of the band limited image signal obtained from the reference original image signal, the parameter of the conversion function (here, the Y value in Table 3), and the original image signal Sorg representing the 200 dpi image to be processed. It is known that the peak frequency of the obtained band-limited image signal is known in advance, and further, the number of conversion functions to be obtained is six. Therefore, first, the frequency response characteristic of the band-limited image signal of the reference original image signal, the peak frequency of the band-limited image signal based on the original image signal of 200 dpi resolution, and the parameter Y of the reference conversion function are expressed by the following equation (12). A response at a frequency corresponding to the peak frequency of the band limited image signal based on the original image signal Sorg is obtained.
R1 = X1 [F1] * Y1 + X2 [F1] * Y2 + X3 [F1] * Y3 + X4 [F1] * Y4 + X5 [F1] * Y5 + X6 [F1] * Y6
R2 = X1 [F2] * Y1 + X2 [F2] * Y2 + X3 [F2] * Y3 + X4 [F2] * Y4 + X5 [F2] * Y5 + X6 [F2] * Y6
R3 = X1 [F3] * Y1 + X2 [F3] * Y2 + X3 [F3] * Y3 + X4 [F3] * Y4 + X5 [F3] * Y5 + X6 [F3] * Y6
R4 = X1 [F4] * Y1 + X2 [F4] * Y2 + X3 [F4] * Y3 + X4 [F4] * Y4 + X5 [F4] * Y5 + X6 [F4] * Y6
R5 = X1 [F5] * Y1 + X2 [F5] * Y2 + X3 [F5] * Y3 + X4 [F5] * Y4 + X5 [F5] * Y5 + X6 [F5] * Y6
R6 = X1 [F6] * Y1 + X2 [F6] * Y2 + X3 [F6] * Y3 + X4 [F6] * Y4 + X5 [F6] * Y5 + X6 [F6] * Y6
(12)
However, R1 to R6: Response at peak frequency (200 dpi image)
X1 [F1] to X6 [F6]: Response at the frequency of F1 to F6 of the band limited image signal created based on the reference original image signal
Y1 to Y6: Standard conversion function parameters

When the responses R1 to R2 are obtained in this way, the parameters A1 to A6 of the conversion function corresponding to the 200 dpi resolution image are obtained by solving the following simultaneous equations (13).
R1 = Z1 [F1] * A1 + Z2 [F1] * A2 + Z3 [F1] * A3 + Z4 [F1] * A4 + Z5 [F1] * A5 + Z6 [F1] * A6
R2 = Z1 [F2] * A1 + Z2 [F2] * A2 + Z3 [F2] * A3 + Z4 [F2] * A4 + Z5 [F2] * A5 + Z6 [F2] * A6
R3 = Z1 [F3] * A1 + Z2 [F3] * A2 + Z3 [F3] * A3 + Z4 [F3] * A4 + Z5 [F3] * A5 + Z6 [F3] * A6
R4 = Z1 [F4] * A1 + Z2 [F4] * A2 + Z3 [F4] * A3 + Z4 [F4] * A4 + Z5 [F4] * A5 + Z6 [F4] * A6
R5 = Z1 [F5] * A1 + Z2 [F5] * A2 + Z3 [F5] * A3 + Z4 [F5] * A4 + Z5 [F5] * A5 + Z6 [F5] * A6
R6 = Z1 [F6] * A1 + Z2 [F6] * A2 + Z3 [F6] * A3 + Z4 [F6] * A4 + Z5 [F6] * A5 + Z6 [F6] * A6
... (13)
However, R1 to R6: Response obtained by the above formula (12)
Z1 [F1] to Z6 [F6]: Responses at the frequencies F1 to F6 of the band limited image signal created based on the 200 dpi resolution original image signal
A1 to A6: Conversion function parameters

  The frequency response of the processed image signal Sproc regardless of the resolution is obtained by performing frequency enhancement processing on the image signal with 200 dpi resolution in the same manner as in the third embodiment using the conversion function thus obtained. The characteristics can be made substantially the same, whereby a processed image signal Sproc having substantially the same frequency response characteristics can be obtained regardless of the resolution of the original image.

  Here, a method of calculating Z1 [F1] to Z6 [F6] in the above equation (13) will be described. The band limited image signal of the reference original image signal Sorg0 having a resolution of 10 lines / mm has the characteristics shown in FIG. In the characteristic shown in FIG. 8, the characteristic of the highest frequency band indicated by a solid line is the result of Fourier transform of the filter coefficient of the one-dimensional filter shown in FIG. 4 and the result of Fourier transform of the filter coefficient of the one-dimensional filter shown in FIG. The frequency response characteristic of the blurred image signal is obtained by multiplication, and further this frequency response characteristic is obtained by subtracting this frequency response characteristic from the frequency response characteristic of the reference original image signal Sorg0 (having a value of 1 over the entire frequency band). Further, for the characteristics of the low frequency band, the frequency response characteristic of the blurred image signal is obtained in the same manner as described above based on the filter coefficient of the filter for obtaining the low resolution image signal corresponding to the frequency band and the filter coefficient for obtaining the blurred image signal. This is obtained by subtracting this from the frequency response characteristic of the blurred image signal in one high frequency band.

Here, the response of the band limited image signal in the highest frequency band of the reference original image signal Sorg0 is represented by a function X1 [Fi] (Fi: frequency), and the response of the band limited image signal when the frequency is sampled by 2048 is X1. Given the function [i], the relationship between Fi and i is
i = 2047 x Fi / fnq (14)
However, fnq: Nyquist frequency 0 ≦ i ≦ 2047
It becomes. This correspondence relationship is shown in FIG. Note that i is an address when the function X1 [i] is handled as a table. In this way, the responses in the entire frequency band of the band-limited image signal are obtained as functions X1 [i] to X6 [i], and these functions X1 [i] to X6 [i] are handled as a table, whereby the above formula ( The values of X1 [F1] to X6 [F6] in 12) and Equation (13) can be easily obtained using Equation (14).

Responses Z1 [F1] to Z6 [F6] are obtained based on the response functions X1 [i] to X6 [i] of all frequency bands obtained in this way. Here, a case where a response Z1 [Fi] at an arbitrary frequency Fi is obtained from the function X1 [i] will be described. The response Z1 [Fi] is a response at the frequency Fi of the band limited image signal of the highest frequency band among the band limited image signals created based on the original image signal of 200 dpi resolution. The Nyquist frequency of the original image signal with 200 dpi resolution is 4 cycles / mm, and the response when the frequency of the response Z1 [Fi] is sampled by 2048 is expressed by the function Z1 [i]. Similar to X1 [i], the relationship is as shown in equation (14). FIG. 30 shows the relationship between the function X1 [i] and the function Z1 [i]. Conceptually, the function X1 [i] and the function Z1 [i] are different functions having different Nyquist frequencies, but the response Z1 [Fi] can be approximately obtained using the function X1 [i]. For example, when obtaining the response Z1 [Fi] of 2 cycle / mm, the above equation (14)
i = 2047 × 2/4 = 1024
Therefore, the value where the address in the function X1 [i] is 1024 is obtained, and it is obtained as the response Z1 [Fi] at 2 cycles / mm.

  As described above, the function X1 [i] to X6 [i] is stored as a table, and the address i is obtained based on the Nyquist frequency of the original image signal to be subjected to frequency enhancement processing and the above equation (14), thereby obtaining the function X1 [i ] To X6 [i] with reference to the table, responses Z1 [F1] to Z6 [F6] can be obtained approximately.

  When the frequency enhancement process is performed on the original image signal Sorg representing the 300 dpi resolution image, based on the frequency response characteristic of the band limited image signal of the reference original image signal and the 300 dpi resolution image signal according to the above equation (12). A response at a frequency corresponding to the peak frequency of the band limited image signal based on the 300 dpi resolution original image signal is obtained from the peak frequency of the band limited image signal and the parameter Y of the reference conversion function. And the parameter of a conversion function can be calculated | required by said Formula (13). At 300 dpi resolution, the Nyquist frequency is 5.9 cycle / mm, which exceeds the Nyquist frequency 5 cycle / mm of the reference original image signal Sorg0. Here, since X1 [F1] to X6 [F6] in the above equation (12) are considered as function values for the range of 0 to 5 cycle / mm, there is no value corresponding to 5.9 cycle / mm. R1 in 12) cannot be obtained. Therefore, in the 300 dpi resolution image, the peak frequency of the band limited image signal obtained from the reference original image signal is used as the peak frequency.

A case where the frequency emphasis process is performed on the original image signal Sorg representing an image with 100 dpi resolution will be described below. Since the Nyquist frequency of 100 dpi resolution is about 2.0 cycle / mm (strictly 1.97 cycle / mm), the number of band-limited image signals is five. Therefore, the above equations (12) and (13) are changed as shown in the following equations (15) and (16), thereby obtaining five parameters of the conversion function.
R1 = X1 [F1] * Y1 + X2 [F1] * Y2 + X3 [F1] * Y3 + X4 [F1] * Y4 + X5 [F1] * Y5 + X6 [F1] * Y6
R2 = X1 [F2] * Y1 + X2 [F2] * Y2 + X3 [F2] * Y3 + X4 [F2] * Y4 + X5 [F2] * Y5 + X6 [F2] * Y6
R3 = X1 [F3] * Y1 + X2 [F3] * Y2 + X3 [F3] * Y3 + X4 [F3] * Y4 + X5 [F3] * Y5 + X6 [F3] * Y6
R4 = X1 [F4] * Y1 + X2 [F4] * Y2 + X3 [F4] * Y3 + X4 [F4] * Y4 + X5 [F4] * Y5 + X6 [F4] * Y6
R5 = X1 [F5] * Y1 + X2 [F5] * Y2 + X3 [F5] * Y3 + X4 [F5] * Y4 + X5 [F5] * Y5 + X6 [F5] * Y6
... (15)
However, R1 to R5: Response at peak frequency (100 dpi image)
X1 [F1] to X6 [F5]: Response at the frequency of F1 to F5 of the band limited image signal created based on the original image signal
Y1 to Y6: Standard conversion function parameters
R1 = Z1 [F1] * A1 + Z2 [F1] * A2 + Z3 [F1] * A3 + Z4 [F1] * A4 + Z5 [F1] * A5
R2 = Z1 [F2] * A1 + Z2 [F2] * A2 + Z3 [F2] * A3 + Z4 [F2] * A4 + Z5 [F2] * A5
R3 = Z1 [F3] * A1 + Z2 [F3] * A2 + Z3 [F3] * A3 + Z4 [F3] * A4 + Z5 [F3] * A5
R4 = Z1 [F4] * A1 + Z2 [F4] * A2 + Z3 [F4] * A3 + Z4 [F4] * A4 + Z5 [F4] * A5
R5 = Z1 [F5] * A1 + Z2 [F5] * A2 + Z3 [F5] * A3 + Z4 [F5] * A4 + Z5 [F5] * A5
... (16)
However, R1 to R5: Response obtained by the above formula (15)
Z1 [F1] to Z5 [F5]: Responses at the frequencies F1 to F5 of the band limited image signal created based on the original image signal with 100 dpi resolution
A1 to A5: Conversion function parameters

  In the fourth and fourth embodiments, the frequency emphasis process shown in the above equation (2) is performed as the conversion process, but the frequency emphasis process and the dynamic range compression process may be performed simultaneously. Good. Hereinafter, an apparatus for performing this process will be described as a fifth embodiment.

  FIG. 31 is a schematic block diagram showing the configuration of a device that performs frequency enhancement processing and dynamic range compression processing as an example of conversion processing according to the fifth embodiment of the present invention, together with the blurred image signal creation means 2. As shown in FIG. 31, the difference between the original image signal Sorg and the blurred image signal Susk created by the filtering processing means 10 and the interpolation processing means 11 is obtained by the subtractor 21, and the limit of the original image signal is obtained. Band-limited image signals (Sorg-Sus1, Sus1-Sus2, etc.) that are components of the frequency band thus created are created. The band-limited image signal thus obtained is input to the first conversion means 3a and the second conversion means 3b as shown in FIG. 31, and is processed by the converters 22a and 22b of each conversion means. .

  The conversion by the converter 22a of the first conversion means 3a is performed using the conversion function calculated according to the resolution of the original image signal Sorg as described above. Specifically, for example, the conversion function shown in FIGS. 32 and 33 or a combination of these functions is used as a reference conversion function, and the frequency response characteristics of the processed image signal Sproc are substantially the same regardless of the resolution of the original image signal Sorg. Based on the reference conversion function, a conversion function corresponding to the resolution is calculated, and the calculated conversion function is used.

  Here, the conversion function shown in FIG. 32 performs conversion that suppresses a band-limited image signal having a large amplitude, and the degree of suppression of the band-limited image signal having a high frequency band is set to a band limit having a low frequency band. Although it is stronger than the image signal, it takes into account that the amplitude of the high frequency component contained in the edge of the actual radiographic image is smaller than that of the low frequency component. In an actual radiographic image, even a steep edge is not accurately stepped, and the amplitude is often reduced as the frequency component becomes higher. For this reason, it is desirable to suppress the band-limited image signal having a higher frequency from a smaller amplitude in accordance with the amplitude of each frequency component, and this function can realize this.

  The function shown in FIG. 33 is for converting the band-limited image signal so that the band-limited image signal is determined based on the absolute value of the band-limited image signal, and is equal to or less than the absolute value. The more the function, the smaller the absolute value of the converted image signal obtained when the band limited image signal whose absolute value is within a predetermined range near 0 is converted. To do. In other words, each of these functions passes through the origin, the slope of the function is 1 or less regardless of the value processed by the function, and the slope in the vicinity of 0 of the function is a function for processing the low frequency band. It is characterized by being as small as possible. In these functions, when a signal obtained by integrating the converted image signals is added to the original image signal Sorg, the connection between the original image signal Sorg and the added signal, that is, the rise of the signal becomes more natural. There is an effect.

  Similarly, the conversion by the converter 22b of the second conversion means 3b is performed using, for example, the function shown in FIG. 34, the function of FIG. 33 described above, or a combination of the functions of FIG. 34 and FIG. The converted image signals output by the conversion means 3a and 3b are input to the calculators 23a and 23b, respectively. Here, the calculator 23a performs a calculation for creating a signal necessary for the frequency enhancement process, and the calculator 23b performs a calculation for creating a signal necessary for the dynamic range compression process.

  The computing unit 23a performs frequency enhancement processing similar to that in the third embodiment described above. The calculator 23a performs the following processing. First, as described above, the band limited image signals converted by the first conversion means 3a are integrated. When this integrated signal is obtained, the degree of enhancement β corresponding to the value of the original image signal Sorg is further multiplied.

  On the other hand, the calculator 23b performs dynamic range compression processing. The calculator 23b performs the following processing. First, as described above, the band limited image signals converted by the second conversion means 3b are integrated. The obtained integrated signal is subtracted from the original image signal Sorg. Further, the difference signal created by the subtraction is converted by a conversion function, and a dynamic range compression coefficient is obtained.

  The signals obtained by the calculators 23a and 23b are added to the original image signal Sorg by the adding means 28, respectively, and a processed image signal Sproc is obtained.

The above processing is expressed by the following equation (3) similar to that described in the prior art.
Sproc = Sorg + β (Sorg) · Fusm (Sorg, Sus1, Sus2, ... SusN)
+ D (Sorg−Fdrc (Sorg, Sus1, Sus2,... SusN)) (3)
Fusm (Sorg, Sus1, Sus2, ... SusN)
= {Fu1 (Sorg−Sus1) + fu2 (Sus1−Sus2) +
+ Fuk (Susk-1−Susk) +... + FuN (SusN−1−SusN)}
Fdrc (Sorg, Sus1, Sus2, ... SusN)
= {Fd1 (Sorg−Sus1) + fd2 (Sus1−Sus2) +
+ Fdk (Susk-1−Susk) +... + FdN (SusN−1−SusN)}
(However, Sproc: processed image signal
Sorg: Original image signal
Susk (k = 1 to N): Blur image signal
fuk (k = 1 to N): conversion function used for the first conversion process
fdk (k = 1 to N): conversion function used for the second conversion process
β (Sorg): Enhancement coefficient determined based on the original image signal
D (Sorg-Fdrc): determined based on the low frequency component signal
Dynamic range compression factor
(D is a function that converts Sorg-Fdrc)

  Here, in the fifth embodiment shown in FIG. 31, a signal used for frequency emphasis processing and a signal used for dynamic range compression processing are respectively created based on the original image signal Sorg and finally added. However, there is a case where either the frequency enhancement process or the dynamic range compression process is first performed on the original image signal Sorg, and the other one process is performed on the resulting signal. Shall be included. However, in general, the low-density part of the radiographic image is a part with a small imaging dose at the time of imaging, and therefore contains a relatively large amount of noise. On the other hand, in the frequency emphasis process, the density-dependent emphasis process is performed such that the emphasis degree becomes higher as the high-density part increases. Therefore, if the frequency emphasis process is performed directly on the original image signal, the low-density part is not emphasized. The noise is never emphasized. On the other hand, for example, when the frequency enhancement process is performed after the dynamic range compression process is performed on the original image signal, the density of the low density portion is increased by the dynamic range compression process. That is, the part is emphasized in the enhancement process, and noise included in the part is enhanced at the same time. Therefore, each signal is preferably generated based on the original image signal Sorg as in the fifth embodiment shown in FIG. Also in terms of shortening the processing time, it is desirable to perform the two processes in parallel.

  As described above, even when the dynamic range compression process is performed, the conversion function is calculated so that the frequency response characteristics of the processed image signal Sproc are substantially the same according to the resolution of the original image signal Sorg, thereby obtaining the original function. A processed image signal Sproc having substantially the same frequency response characteristics can be obtained regardless of the resolution of the image signal Sorg.

  As described in the third embodiment, not only a set of reference conversion functions corresponding to a resolution of 10 lines / mm is used in the fourth and fifth embodiments, but also on the right side of Table 3. As shown in the figure, conversion functions corresponding to several types of reference resolutions that are frequently used are prepared as a set of reference conversion functions. When an image having a predetermined resolution other than the basic resolution is used as an original image, the original image subjected to this processing is used. The optimal resolution is selected from the set, such as determining whether the predetermined resolution is close to any of the above-mentioned several types of reference resolutions, and selecting the optimal set from the set, and the selected set of reference conversion functions A conversion function corresponding to the resolution of the image to be processed may be calculated using the basic resolution and the predetermined resolution information, and the conversion process may be performed using the calculated conversion function. At this time, as a method of selecting an appropriate basic resolution for a processing target from a plurality of basic resolutions, the method described in JP-A-10-63838 may be used, or may be set by trial and error, or the following Such a method can also be used.

First, a table of definition parameters with basic resolutions (a, b, c; a <b <c: c is the highest resolution, and in the above example, a = 5, b = 6.7, c = 10) as shown in Table 3 above. If the original image resolution Rs to be processed is different from the basic resolution (a, b, c; a <b <c: c is the highest resolution), the basic resolutions a, b, c The closest basic resolution definition parameter is selected according to the following rules.
When Rs> b + (c−b) / 2, c parameter b + (c−b) / 2 ≧ Rs> a + (ba−2) When b parameter b + (c−b) / Parameter for a when 2 ≧ Rs By using this method, the definition parameter of the conversion function can be easily determined.

  Further, when the system has a plurality of basic resolutions and parameters for each basic resolution as described above, one basic resolution and basic resolution parameter are selected from them, and the selected basic resolution and The parameters and the resolution information of the image to be processed may be stored in association with the image to be processed. Thereby, when re-outputting the image to the film or the monitor, the definition parameter corresponding to the image output last time is read from the storage device, and the processed image signal proc is generated using the read definition parameter, An image with the same image quality as the image output last time can be obtained. Further, since the process of determining the definition parameter can be omitted, an image is output in a short time.

  When performing the frequency enhancement processing shown in the third and fourth embodiments or the dynamic range compression processing shown in the fifth embodiment, a processed image signal obtained by processing a predetermined resolution image signal. The original image signal (such as the reference original image signal Sorg0 in the above example) is processed at a frequency equal to or lower than 1/5 (more preferably 1/2) of the Nyquist frequency that defines the predetermined resolution. The conversion function is preferably calculated so as to substantially match the characteristics of the image represented by the processed image signal Sproc obtained in this way. Hereinafter, this point will be described with reference to FIG. 35 showing the relationship between the emphasis characteristic and the response of the image represented by the processed image signal.

  For example, in the case of a system in which there are 5 lines / mm, 6.7 lines / mm, and 10 lines / mm as basic resolutions, and a conversion function group is defined for each, for example, the resolution of an image to be processed Is 4 lines / mm, the closest basic resolution is 5 lines / mm, so the conversion function group for 4 lines / mm is used as described above using the conversion function group for 5 lines / mm. The group will be calculated.

  Here, when each band limited image signal is obtained, the image signal is reduced by ½ unit. Therefore, in the image signal for 4 lines / mm, the highest frequency band signal among the band limited image signals. The characteristic width (for example, a substantially flat portion with a response of 0.8 or more) is at least about 1.0 cycle / mm to 2.0 cycle / mm, and when performing a conversion process in which the gain of each band limited image signal is controlled by a conversion function For example, if you try to match the characteristics near 1.0 cycle / mm to the basic resolution characteristics, you can not raise the response near 2.0 cycle / mm, and it will not match the basic resolution characteristics near 2.0 cycle / mm. If you try to match the characteristics in the vicinity of cycle / mm with the basic resolution characteristics, the response in the vicinity of 1.0 cycle / mm will rise, so it will not match the basic resolution characteristics in the vicinity of 1.0 cycle / mm. 2.0cycle / mm It is difficult to fine-tune the response of (can not).

  Therefore, as a conversion function group for a basic resolution of 5 lines / mm, as shown in FIG. 35A, a conversion function group (parameters) having characteristics that emphasize from the relatively low frequency side to the relatively high frequency side, as shown in FIG. When the conversion function group for 4 lines / mm is calculated, the conversion function for the highest frequency band signal is used as the conversion function for the low frequency band signal excluding the highest frequency band signal. As shown in FIG. 35B, the response of the processed image signal Sproc based on the calculated conversion function group can be set to 1.0 cycle. Even in the vicinity of / mm to 2.0 cycle / mm, the basic resolution characteristics can be substantially matched.

  However, as a conversion function group for a basic resolution of 5 lines / mm, as shown in FIG. 35C, characteristics that have a sudden characteristic change in a frequency band close to the Nyquist frequency, such as emphasizing only a relatively high frequency component. When the conversion function group is set as follows, there is a problem that it is difficult to fine-tune the response between 1.0 cycle / mm and 2.0 cycle / mm as described above, and thus the processed image signal Sproc is It is impossible to make the response coincide with the basic resolution characteristics over the entire frequency band. In this case, as shown in FIG. 35D, importance is placed on the response on the low frequency side that has a large visual influence (the degree of influence on the image quality is large). For example, the Nyquist frequency (2.0 cycle / mm in this example) It is preferable to make the response of the processed image signal Sproc coincide with the basic resolution characteristic at a frequency equal to or lower than 1/5. In addition, better results can be obtained by focusing on the low frequency band component of 1/2 or less of the Nyquist frequency so as to match the basic resolution characteristics.

  In the third to fifth embodiments, the conversion function is calculated according to the resolution of the input original image signal Sorg. However, the conversion function is converted based on the characteristic information regarding the response of the input original image signal Sorg. The function may be corrected. That is, as a device for obtaining the original image signal Sorg, there are a semiconductor sensor and the like in addition to a device that reads a radiation image from a stimulable phosphor sheet, and the response of the obtained original image signal Sorg varies depending on the device. It is. For this reason, a corrected image signal having a constant frequency response characteristic is obtained regardless of the apparatus that acquires the original image signal Sorg by correcting the conversion function based on the characteristic information regarding the response of the original image signal Sorg according to the apparatus. Sproc can be obtained. For example, when the response at 2 cycles / mm of the original image signal Sorg obtained in a certain apparatus is 40% inferior to the reference response, that fact is input to the image processing apparatus according to the present invention as characteristic information, In the image processing apparatus, the conversion function is corrected so that the response of 2 cycle / mm becomes the reference response, and the processed image signal Sproc is obtained by the corrected conversion function.

  Furthermore, in the fourth and fifth embodiments, the conversion function is applied to the band limited image signal using the conversion function as a nonlinear function. However, the conversion function is not limited to the nonlinear function, and the conversion function is a linear function. Or a constant.

  In the third to fifth embodiments, the blurred image signal is obtained from the original image signal by filtering and interpolation expansion, and the band limited image signal is created from the original image signal and the blurred image signal. Without limitation, for example, the original image signal is converted to multiple resolutions using a method such as wavelet transform or Laplacian pyramid, and a blurred image signal is created from the converted image signal for each resolution to create a band-limited image signal May be.

1 is a schematic block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention. The figure which shows the state which carries out wavelet transform of the original image signal and encodes it for every hierarchy Schematic block diagram showing a configuration of a blurred image signal creating unit in the first embodiment The figure which shows an example of the filter used for filtering processing Diagram showing details of low-resolution image signal creation processing The figure which shows an example of the filter used for interpolation expansion processing The figure which shows an example of a frequency emphasis processing apparatus Diagram showing frequency response characteristics of band-limited image signal (reference resolution) Diagram showing frequency response characteristics of band-limited image signal (1/2 resolution) Diagram showing frequency response characteristics of band-limited image signal (1/4 resolution) Diagram showing frequency response characteristics of band-limited image signal (1/8 resolution) Diagram showing frequency response characteristics of band-limited image signal (1/16 resolution) The flowchart which shows operation | movement of 1st Embodiment. The figure which shows the correspondence of six blur image signals obtained from a reference | standard resolution image, and the low resolution blur image signal obtained from the low resolution image from the lowest resolution to the frequency band nearest to predetermined resolution. The figure which shows the correspondence of the 6 band limitation image signals obtained from a reference | standard resolution image, and the low resolution zone limitation image signal obtained from the original image signal and low resolution blur image signal based on each resolution image The figure which shows the correspondence of the conversion function in a reference | standard resolution image and each resolution image Schematic block diagram showing a configuration of a blurred image signal creating unit in the second embodiment The figure which shows the correspondence of each band limitation image signal in the image of a standard resolution, and an enlarged image The figure which shows the correspondence of each conversion function in the standard resolution image and the enlarged image The schematic block diagram which shows the structure of the image processing apparatus by the 3rd Embodiment of this invention. Diagram showing frequency response characteristics of processed image signal The figure which shows the frequency response characteristic of a band-limited image signal (300dpi) The figure which shows the frequency response characteristic of the processed image signal obtained by 3rd Embodiment The flowchart which shows the process of 3rd Embodiment Diagram showing conversion function selection status Diagram showing the number of band-limited image signals according to resolution Schematic (200 dpi) showing frequency response characteristics of band limited image signal Diagram showing frequency response characteristics of processed image signal Diagram showing correspondence between response X1 [F1] and function X1 [i] Diagram showing the relationship between function X1 [i] and function Z1 [i] The figure which shows an example of a dynamic range compression processing apparatus The figure showing an example of the conversion function in a 1st conversion means The figure showing the other example of the conversion function in a 1st conversion means The figure showing an example of the conversion function in a 2nd conversion means The figure which shows the relationship between an emphasis characteristic and the response of the image which a processed image signal represents

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Blurred image signal preparation means 3 Frequency emphasis processing means, conversion processing means 4 Parameter setting means 5 Enlargement ratio input means 6 Enlarging / reducing means
10 Filtering means
11, 31 Interpolation processing means
21 Subtractor
22 Converter
23 Calculator
30 Decryption means
40 Conversion function calculation means
50 Resolution information input means

Claims (21)

A plurality of band-limited image signals having mutually different band-limiting characteristics are created from the input image signal, and a conversion process is performed based on a plurality of conversion functions set corresponding to each band-limited image signal. In an image processing method of obtaining a signal and performing a predetermined process on the input image signal based on the converted image signal to obtain a processed image signal,
Preparing a reference conversion function corresponding to each of the band-limited image signals for the reference resolution image, which is used when the conversion processing is performed on the reference resolution image having the reference resolution;
When an image signal representing a processing image having a resolution different from the reference resolution is input, a response of the processed image signal for the processing image based on a difference between the resolution of the processing image and the reference resolution However, the reference conversion function is corrected so as to be substantially the same as the response of the processed image signal for the reference resolution image, and the function obtained by this correction is set as the conversion function for the processing image. ,
An image processing method, wherein the resolution represents a pixel density.   The characteristic of the image represented by the processed image signal obtained by processing the image signal representing the processing image is corrected at least one of the Nyquist frequencies corresponding to the resolution of the processing image. The frequency range of / 5 or less is performed so as to substantially match the characteristics of the image represented by the processed image signal obtained by processing the image signal representing the reference resolution image. The image processing method as described.   The image processing method according to claim 1, wherein information regarding the resolution of the processing image is acquired, and a conversion function for the processing image is set based on the information.   Creating a plurality of blurred image signals having different frequency response characteristics based on the inputted image signal, and creating the band-limited image signal based on the plurality of blurred image signals and the inputted image signal. The image processing method according to any one of claims 1 to 3, wherein the image processing method is characterized.   The image processing method according to claim 1, wherein the reference conversion function is a non-linear function.   The image processing method according to claim 1, wherein the predetermined process is a frequency enhancement process.   The image processing method according to claim 1, wherein the predetermined process is a dynamic range compression process. An image signal generating means for generating a plurality of band limited image signals having mutually different band limiting characteristics from the input image signal, and a plurality of conversion functions corresponding to each of the band limited image signals are set, and the set conversion function And a conversion processing means for obtaining a converted image signal by performing a conversion process based on the conversion image signal and performing a predetermined process on the input image signal based on the converted image signal. In the processing device,
The conversion processing means holds a reference conversion function corresponding to each of the band-limited image signals for the reference resolution image, which is used when the conversion processing is performed on a reference resolution image having a reference resolution, and the reference resolution When an image signal representing a processing image having a resolution different from that of the processing image is input to the image signal creating means, a processed image for the processing image is based on the difference between the resolution of the processing image and the reference resolution. The reference conversion function is corrected so that the response of the signal is substantially the same as the response of the processed image signal for the reference resolution image, and the function obtained by this correction is set as the conversion function for the processing image. Is,
An image processing apparatus, wherein the resolution represents a pixel density.   The characteristic of the image represented by the processed image signal obtained by the conversion processing means correcting the reference conversion function and processing the image signal representing the processing image depends on the resolution of the processing image. Further, in a frequency range of at least 1/5 or less of the Nyquist frequency, it is performed so as to substantially match the characteristics of the image represented by the processed image signal obtained by processing the image signal representing the reference resolution image. The image processing apparatus according to claim 8.   Resolution information acquisition means for acquiring information related to the resolution of the processing image is further provided, and the conversion processing means sets a conversion function for the processing image based on the information acquired by the resolution information acquisition means. The image processing apparatus according to claim 8, wherein the image processing apparatus is an image processing apparatus.   The image signal generating means generates a plurality of blurred image signals having different frequency response characteristics based on the input image signal, and the band limitation based on the plurality of blurred image signals and the input image signal. The image processing apparatus according to claim 8, wherein the image processing apparatus creates an image signal.   The image processing apparatus according to claim 8, wherein the reference conversion function is a nonlinear function.   The image processing apparatus according to claim 8, wherein the predetermined process is a frequency enhancement process.   The image processing apparatus according to claim 8, wherein the predetermined process is a dynamic range compression process. A procedure for creating a plurality of band-limited image signals having different band-limiting characteristics from the input image signal and a conversion process based on a plurality of conversion functions set corresponding to each band-limited image signal A program for causing a computer to execute an image processing method including a procedure of obtaining a converted image signal and performing a predetermined process on the input image signal to obtain a processed image signal based on the converted image signal In a computer-readable recording medium that records
A procedure for preparing a reference conversion function corresponding to each of the band-limited image signals for the reference resolution image, which is used when the conversion process is performed on the reference resolution image having the reference resolution;
When an image signal representing a processing image having a resolution different from the reference resolution is input, a response of the processed image signal for the processing image based on a difference between the resolution of the processing image and the reference resolution However, there is a procedure for correcting the reference conversion function so as to be substantially the same as the response of the processed image signal for the reference resolution image, and setting the function obtained by this correction as the conversion function for the processing image. And
The computer-readable recording medium, wherein the resolution represents a pixel density.   The procedure for setting the conversion function includes correcting the reference conversion function, and determining the characteristics of the image represented by the processed image signal obtained by processing the image signal representing the processing image. In a frequency range at least 1/5 or less of the Nyquist frequency corresponding to the resolution, it is performed so as to substantially match the characteristics of the image represented by the processed image signal obtained by processing the image signal representing the reference resolution image. The computer-readable recording medium according to claim 15, wherein the recording medium is a procedure.   The method further comprises a step of obtaining information on the resolution of the processing image, and the step of setting the conversion function is a step of setting a conversion function for the processing image based on the information. The computer-readable recording medium according to claim 15 or 16.   The procedure of creating the plurality of image signals is to create a plurality of blurred image signals having different frequency response characteristics based on the input image signal, and based on the plurality of blurred image signals and the input image signal. The computer-readable recording medium according to any one of claims 15 to 17, wherein the band-limited image signal is generated by a procedure.   The computer-readable recording medium according to claim 15, wherein the reference conversion function is a nonlinear function. The computer-readable recording medium according to claim 15, wherein the predetermined process is a frequency enhancement process.   20. The computer-readable recording medium according to claim 15, wherein the predetermined process is a dynamic range compression process.

Images