JP4013491B2 - Apparatus, method and program for image processing - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to image processing, and in particular to enlargement and reduction of digital images.
[0002]
[Prior art]
The digital image data is subjected to resolution conversion for enlargement or reduction. Bilinear interpolation and bicubic interpolation are generally used. Various methods for effectively performing resolution conversion have been proposed. For example, in Japanese Patent Application Laid-Open No. 5-73675, the characteristics of a filter are determined from load factors before and after scaling, and scaling factors.
[0003]
[Problems to be solved by the invention]
When enlarging / reducing a digital image, using commonly known interpolation methods such as bilinear interpolation and bicubic interpolation, jaggies and edge blurring occur at the edge of the image. In many cases, problems occur in the processing result, such as when there is no edge at the time of reduction. Therefore, it is desirable not to impair edge information in the image at the time of enlargement / reduction.
In order to avoid harming the edge information, the method described in Japanese Patent Laid-Open No. 5-328106 detects an edge, executes one of a plurality of smoothing filters for the edge, and performs interpolation. . In the method described in Japanese Patent Laid-Open No. 8-263644, an edge (contour vector) is detected, subjected to scaling and smoothing, and then subjected to interpolation. The weighting of the average value calculation increases as the magnification ratio increases. However, it is desirable that the processing results be further improved.
[0004]
In addition, when resolution conversion is performed on an image having a channel other than RGB color information (for example, a transparency channel), the other channel is also subjected to the same processing as the RGB channel, and a boundary line between the transparent portion and the non-transparent portion. In some cases, problems may occur in the processing results such as jaggies. Therefore, when there is a channel other than RGB, it is desirable to be able to perform optimal resolution conversion in consideration of that channel.
[0005]
A first object of the present invention is not to impair edge information in an image when the image is enlarged or reduced.
A second object of the present invention is to perform optimum processing in consideration of a channel other than RGB when enlarging / reducing an image.
[0007]
[Means for Solving the Problems]
Further, engagement Ru images processing device according to the present invention includes an image input means for inputting image data, when the image data input by the image input means has another channel other than RGB, the resolution with respect to other channels First resolution conversion means for performing conversion, threshold value determination means for determining a threshold value from each pixel value in the image after the other channel processing by the first resolution conversion means, an image size, and a scaling factor, and a threshold value Filter means for filtering input image data using a variable weighting filter using a threshold value determined by the determination means, and resolution conversion for the image data filtered by the filter means using an interpolation function Second resolution conversion means. As described above, when there is a channel other than RGB, first, resolution conversion is performed on the other channel, and the value after resolution conversion is added to the determination material to determine the threshold value of the RGB channel. In this image processing apparatus, preferably, the threshold value determining means determines a plurality of threshold values from each pixel value in the image after the other channel processing, the image size, and the scaling factor. Then, the filter means performs a filtering process on the input image data using a plurality of different variable weighting filters each using the threshold value determined by the threshold value determination means.
[0009]
Further, the engaging Ru images processing method according to the present invention, image data is input, when the input image data has other channels other than RGB, performs resolution conversion for the other channels, other post channel processing The threshold value is determined from each pixel value in the image, the image size, and the scaling factor. Next, the input image data is filtered using a variable weighting filter that uses the determined threshold value, and resolution conversion is performed on the filtered image data using an interpolation function. Preferably, a plurality of threshold values are determined from each pixel value in the image after the other channel processing, the image size, and the scaling factor. Next, the input image data is filtered using a plurality of different variable weighting filters each using the determined threshold value, and resolution conversion is performed on the filtered image data using an interpolation function.
[0011]
The image processing program for engaging Turkey computer execute the present invention includes the steps of inputting image data, when the input image data has other channels other than RGB, a step of performing resolution conversion on the other channel, A step of determining a threshold value from each pixel value in the image data after the channel processing, the size of the input image data, and a scaling factor, and a variable weighting filter using the determined threshold value The method includes a step of filtering input image data and a step of performing resolution conversion on the image data subjected to the filter processing using an interpolation function. Preferably, the plurality of threshold values are determined from each pixel value in the image after the channel processing, the size of the image, and the scaling factor. Then, the input image data is filtered using a plurality of different variable weighting filters each using the determined threshold value, and resolution conversion is performed on the filtered image data using an interpolation function.
In this image processing program, for example, when determining the threshold value, the RGB channel is divided into a plurality of areas so as to correspond to the processed pixels of other channels, and the threshold value is determined for each area. To decide.
This image processing program preferably further comprises a step of extracting only edge information of other channels. For each of the RGB channels, at least one filter is applied only to the portion corresponding to the multi-channel edge.
The computer-readable recording medium according to the present invention records any of the above-described image processing programs.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.
FIG. 1 shows a block diagram of a configuration in an embodiment of the present invention. Image data read by an image input device 100 typified by a digital camera or the like is sent to an image processing device 200 typified by a computer or the like, where an appropriate resolution conversion process is performed on the image data. It is sent to an image output device 300 represented by a printer, a display, etc., and output. The image processing apparatus 200 has a configuration as a normal computer. That is, the CPU 202 has a ROM 204 storing a program, a RAM 206 as a work area, a flexible disk device 208 and a hard disk device 210 as storage devices, a keyboard 212 and a mouse 214 as input devices, a display device 216, and an external device. I / O interface 218 for connection of
[0013]
In resolution conversion for enlarging or reducing an image, it is desirable not to lose edge information of the image. For this reason, in resolution conversion, a plurality of threshold values are dynamically determined based on the size and magnification of image data, and then a plurality of variable weighting filters with different threshold values are applied. Thereafter, an interpolation function (interpolation algorithm) is applied. The threshold value is determined from the image size and the scaling factor, but considering the amount of information of the original image, the optimum threshold value is calculated from the size of the image after processing to increase the filter effect. Here, when two threshold values are used, for example, the filter based on the first threshold is applied only to the edge portion in the image, and the filter based on the second threshold is applied to the entire image data. Apply to. As a filter, at the time of reduction, the weighted average value filter is used to blur and thicken the image. In addition, a weighted sharpening filter is used during enlargement. By applying the filter to the edge part and the whole in two stages, in the image data after the interpolation, the defect of the edge and the step at the boundary part between the edge and the part that is not so conspicuous are made inconspicuous.
[0014]
In addition, when there is image data (α channel) other than RGB in enlargement or reduction of an image, first, resolution conversion is performed on the data of other channels. Next, the value after the resolution conversion is added to the determination material for the RGB channel data, and the threshold value of the RGB channel is determined. As described above, when there is a channel other than RGB, first, resolution conversion is performed on the channel, and a threshold value for each pixel value in the image data after the channel processing is determined. Edge defects such as jaggies are eliminated on the boundary between the transparent portion and the non-transparent portion of the image.
[0015]
Here, when determining one of the threshold values, the RGB image data is divided into a plurality of regions (for example, rectangular regions) for each of the RGB channels so as to correspond to the processed pixels of the other channels, A threshold is determined for each region.
[0016]
Also, only edge information of other channels is extracted, and the value after resolution conversion is added to the judgment material. That is, for each RGB channel, at least one filter is applied only to the portion corresponding to the edge.
[0017]
FIG. 2 shows a flowchart of resolution conversion. In the resolution conversion, when digital image data is input, it is first determined whether or not there is data of an α channel (for example, a transparency channel) other than RGB (S10).
If there is alpha channel data, the following processing is performed. First, two threshold values TH1 and TH2 for the α channel are determined (S12). Next, using these threshold values, an edge is detected for the data of the α channel (S14). An edge is a portion where data such as a step of a frame is largely converted. First, a weighting filter having a first threshold value TH1 is applied to edge data in the α channel data (S16). Next, a weighting filter of the second threshold value TH2 is applied to the data of the entire α channel (S18). Then, resolution conversion is performed on the α channel using an interpolation function (for example, bilinear interpolation method or bicubic interpolation method) (S20). Thereby, α-channel processed image data is obtained.
[0018]
Next, two threshold values TH1 and TH2 for the RGB channels are determined (S22). Next, the RGB data is converted into HSL color space data (S24).
Next, an edge is detected from the L data (S26). Then, a weighting filter of the first threshold value TH1 is applied to each of the RGB channels for the edge portion (S28).
If there is an α channel (YES in S30), the filter of the third threshold TH3 is applied to the pixels of each RGB channel corresponding to each pixel of the α channel processed image data obtained in step S20 (S32). . Next, a weighting filter of the second threshold value TH2 is applied to the entire RGB channels (S34).
Next, resolution conversion is performed on each of the RGB channels with an interpolation function (for example, bilinear interpolation or bicubic interpolation) (S36). Thus, the α channel processed image obtained in step S20 and the RGB channel processed image obtained in step S36 are output.
[0019]
3 shows an example of a 3 × 3 weighting filter, and FIG. 4 shows an example of a 3 × 3 sharpening filter. Here, W and a are functions of the threshold value TH. In these filters, the factor of normalization is actually multiplied. For example, in the case of FIG. 3, the normalization factor is 1 / (8 / W + 1).
[0020]
【The invention's effect】
Since the threshold value is determined from the image size and the magnification, the optimum threshold value is calculated from the processed image size in consideration of the information amount of the original image, and the effect of the filter is improved.
Since the filter is sequentially applied to the edge portion and the entire image data, defects at the edge are eliminated in the interpolated data, and the step at the boundary portion between the edge and the other portion is made inconspicuous.
When there is a channel other than RGB, first, resolution conversion is performed on the channel, and a threshold value for each pixel value in the image after the channel processing is determined. As a result, for example, an edge defect such as a jaggy on the boundary line between the transparent portion and the non-transparent portion is eliminated with respect to the frame image or the character image.
[Brief description of the drawings]
FIG. 1 is a block diagram of a system including an image processing apparatus. FIG. 2 is a resolution conversion flowchart. FIG. 3 is an example of a weighting filter. FIG. 4 is an example of a sharpening filter.
100 image input device, 200 image processing device, 202 CPU 202, 208 flexible disk device, 210 hard disk device, 300 image output device.

Claims (6)

Image input means for inputting image data;
First resolution conversion means for performing resolution conversion on other channels when the image data input by the image input means has channels other than RGB;
Threshold value determining means for determining a threshold value from each pixel value in the image after the other channel processing by the first resolution converting means, the image size, and the scaling factor;
Filter means for filtering input image data using a variable weighting filter using a threshold value determined by the threshold value determination means;
An image processing apparatus comprising: second resolution conversion means for performing resolution conversion using an interpolation function on the image data filtered by the filter means. When the input image data has channels other than RGB, resolution conversion is performed on the other channels,
The threshold value is determined from each pixel value in the image after other channel processing, the image size, and the scaling factor,
Filter the input image data using a variable weighting filter using the determined threshold,
An image processing method for performing resolution conversion on an image data subjected to filter processing using an interpolation function. Inputting an image;
When the input image data has a channel other than RGB, performing resolution conversion on the other channel;
Determining a threshold value from each pixel value in the image data after the channel processing, the size of the input image data, and a scaling factor;
An image processing program for causing a computer to execute a filtering process of input image data using a variable weighting filter using a determined threshold value. In determining the threshold, so as to correspond to the pixel after processing of other channels, according to claim 3 which is divided into a plurality of regions with respect to each of RGB channels, to determine a threshold for each area Image processing program. Furthermore, it has a step of extracting only edge information of other channels,
The image processing program according to claim 3 or 4 , wherein at least one filter is applied to each of the RGB channels only at a location corresponding to an edge of another channel. A computer-readable recording medium for recording the image processing program according to claim 3 .

Images