JP2006135566A - Image processing apparatus and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method, which obtain a reduced image with high quality even when reduction processing at any magnification is carried out, in a configuration provided with a data processing section for realizing reduction of image data by means of hardware. <P>SOLUTION: A magnification block 8 and a second reduction block 10 connected in series magnify or reduce an image comprising YUV data transferred from a Bayer interpolation block 5 or read from a memory 3 and transferred via a switch 7 into an image of a target size in response to a requested magnification. A combined magnification in response to the requested magnification is set to the magnification block 8 and the second reduction block 10. While attaining the magnification or reduction by the requested magnification, reduction processing at a reduction rate at which image deterioration is always caused can be carried out by the reduction processing section. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing apparatus and an image processing method suitable for use in a digital camera.

  2. Description of the Related Art Conventionally, for example, in a digital camera that records digital image data acquired by photographing on a recording medium, multiple stages of image processing are performed in the process of finally obtaining image data to be recorded. For example, Bayer interpolation processing for converting Bayer data temporarily stored in a memory at the time of photographing, that is, image data including pixels of color information corresponding to a color filter of a primary color Bayer array into YUV data including RGB data for each pixel, after conversion Image processing such as scaling processing that realizes a digital zoom function or the like by increasing or decreasing the number of pixels of YUV data. Also, in such image processing, the data of the whole image (Bayer data or YUV data) is not used as a processing unit, but data for a plurality of lines called a belt-like belt obtained by dividing the entire screen A in the vertical direction is used as a processing unit. Then, data is read out from the working memory in order for each belt and processed (see FIG. 2A). Note that this belt may overlap the previous belt depending on the processing content (see FIG. 2B).

Further, when the processing is performed by hardware for the purpose of speeding up the scaling process, it is performed by the same algorithm in a single processing block. For example, a new pixel is generated by interpolation between two or several points in the original image, or a new pixel line is generated by interpolation between two or several lines of pixel lines in the belt input to the processing block. Many methods enable enlargement or reduction by a single processing block (see Patent Document 1 below).
JP 2003-18610 A

  However, when the enlargement and reduction are performed using the same algorithm as described above, apart from the enlargement process, the reduction process is a pixel thinning process when the process is performed at a certain magnification or lower. There was a problem that a drop occurred. For example, in the case where the reduction in the vertical direction is performed by the interpolation between the two lines described above, the image quality deteriorates when the magnification is less than 1/2.

  The present invention has been made in view of such a conventional problem, and in a configuration including a data processing unit for realizing reduction of image data by hardware, no matter what magnification reduction processing is performed, An object of the present invention is to provide an image processing apparatus and an image processing method capable of obtaining a reduced image of quality.

  In order to solve the above-mentioned problem, in the invention of claim 1, an enlargement processing unit for enlarging the image data of the transferred original image, and an enlargement processing unit are connected in series to the subsequent stage of the enlargement processing unit. And a reduction processing unit that performs reduction processing on the image data after the enlargement processing, and a magnification setting unit that sets a combination magnification according to the requested magnification in the enlargement processing unit and the reduction processing unit. .

  In such a configuration, by performing a continuous enlargement process and a reduction process, the image data of the original image is scaled at a required magnification. Therefore, in the reduction process, image degradation always occurs. Reduction processing at a reduction ratio can be performed.

  According to the invention of claim 2, the reduction processing unit reduces the image data after the enlargement processing by the enlargement processing unit based on an image processing algorithm different from the enlargement processing.

  In such a configuration, the image data can always be performed based on the image processing algorithm most suitable for the reduction process.

  According to a third aspect of the present invention, a plurality of the reduction processing units are connected in parallel to the subsequent stage of the enlargement processing unit, and the magnification setting unit sets different reduction ratios to the plurality of reduction processing units, respectively. It was supposed to be.

  In such a configuration, reduced images of a plurality of sizes can be obtained simultaneously by a single image data transfer operation.

  According to a fourth aspect of the present invention, the reduction processing unit outputs the image data after the reduction process to a compression unit that compresses and encodes the image data.

  In such a configuration, when the image data of the original image is compressed and encoded, the reduction processing unit can be made to perform reduction processing at a reduction rate that always causes image deterioration.

  According to a fifth aspect of the present invention, transfer means for transferring the image data of the original image to the enlargement processing unit in units of processing consisting of a plurality of lines, and the number of output lines of image data in the reduction processing unit And a line number control means for controlling the number of transfer lines of image data transferred to the enlargement processing unit by the transfer means for each processing unit based on the requested magnification. It was assumed.

  In such a configuration, the number of transfer lines of image data transferred to the enlargement processing unit is adjusted for each processing unit in accordance with the requested magnification regardless of the requested magnification. The number of output lines of image data in the enlargement processing unit can be controlled to the set number of output lines at all times.

  According to a sixth aspect of the present invention, the line number control means includes the requested reduction ratio and the requested reduction ratio to which the transfer means should transfer image data in an image based on the image data. Based on the upper end position of the original image area according to the above and the coordinate positions of the upper end and the lower end of each processing unit according to the number of output lines in the image having a size reduced by the requested reduction ratio. The number of transfer lines of image data transferred to the enlargement processing unit is controlled by the means.

  In the invention of claim 7, in the image processing method for scaling the image data at the requested magnification, a step of setting a combination of an enlargement ratio and a reduction ratio according to the requested magnification. The method includes a step of enlarging the image data at the set enlargement rate and a step of reducing the image data after the enlargement processing at the set reduction rate.

  In such a method, by performing a continuous enlargement process and a reduction process, the image data of the original image is scaled at a required magnification. Therefore, in the reduction process, image degradation always occurs. Reduction processing at a reduction ratio can be performed.

  In the invention of claim 8, the enlargement processing unit for enlarging the image data of the transferred original image is connected in series with the subsequent stage of the enlargement processing unit, and after the enlargement processing by the enlargement processing unit A computer having an image processing apparatus including a reduction processing unit that reduces the image data based on an image processing algorithm different from the enlargement processing, and the enlargement processing unit with a combination ratio according to a requested reduction rate. A program for causing the reduction processing unit to function as a magnification setting unit is provided.

  As described above, in the first and eighth aspects of the present invention, since the enlargement / reduction at the magnification required for the image data of the original image is performed through the continuous enlargement process and reduction process, Therefore, it is possible to perform reduction processing at a reduction rate that always causes image degradation. Therefore, in a configuration including a data processing unit for realizing image data reduction by hardware, a high-quality reduced image can be obtained regardless of the reduction processing at any magnification.

  In the invention of claim 2, image data can always be performed based on an image processing algorithm most suitable for reduction processing. Therefore, it is possible to obtain a reduced image with higher quality.

  According to a third aspect of the present invention, reduced images of a plurality of sizes can be obtained simultaneously by a single image data transfer operation. Therefore, a reduced image having a plurality of sizes can be obtained efficiently.

  In the fourth aspect of the invention, when the image data of the original image is reduced and compressed and encoded, the reduction processing unit can always perform a reduction process at a reduction rate that causes image degradation. it can. Therefore, it is possible to obtain high-quality compressed image data.

  According to the fifth and sixth aspects of the invention, the number of transfer lines of image data transferred to the enlargement processing unit is determined according to the requested magnification regardless of the requested magnification. By adjusting each time, the number of output lines of the image data in the enlargement processing unit can be controlled to the set number of output lines at all times. Therefore, even if another processing block in which the number of input lines of image data is limited exists on the subsequent stage side of the reduction processing unit, the simple circuit configuration can be used and various required magnifications can be accommodated. It becomes possible.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an image processing apparatus 1 according to the present invention. This image processing apparatus 1 is incorporated in a digital camera having a digital zoom function, and is set with interpolation processing for generating YUV data from Bayer data output from an image sensor such as a CCD, and the number of pixels of the generated YUV data. This is for performing a scaling process for increasing / decreasing at a different magnification, and has the following configuration. Note that the scaling process is a process for recording an image having a size smaller than the number of pixels of the image sensor, for example, and for enlarging the image using a digital zoom function of a digital camera.

  In the present embodiment, the control unit 2 controls the entire operation of the digital camera, and includes a CPU, a program ROM, a RAM, a peripheral circuit (not shown) including an input / output interface, and the like. Then, by controlling each part of the image processing apparatus 1 based on the program stored in the program ROM, it functions as the line number setting means, magnification setting means, line number control means, and magnification control means of the present invention.

  The memory 3 is a DRAM or the like in which Bayer data output from a CCD or the like and YUV data are temporarily stored, and is also a transfer destination to which a JPEG encoding result is output, and is sufficient for temporarily storing each data. Have a large capacity.

  The first DMA 4 is a data transfer unit that reads image data in a Bayer data state from the memory 3 in a predetermined processing unit and transfers the image data to the Bayer interpolation block 5 in accordance with an instruction from the control unit 2. The second DMA 6 is a data transfer unit that transfers image data in the YUV data state from the memory 3 to the enlarged block 8 via the read switch 7 in a predetermined processing unit. In response to an instruction from the control unit 2, the switch 7 converts the image data input to the enlargement block 8 into YUV data output from the Bayer interpolation block 5 and YUV data on the memory 3 transferred from the second DMA 6. Switch.

  The image data read from the memory 3 by the first DMA 4 and the second DMA 6 (both are the transfer means of the present invention), the entire image A is vertically set to a predetermined size as shown in FIG. The image data B is the image data of the divided strip-like rectangular portion, and is composed of pixel data for a plurality of lines. This is called a belt. Each belt B (0, 1, 2,...) May overlap data with the preceding belt depending on the processing as shown in FIG. In the first DMA 4 and the second DMA 6, the belt data read from the memory 3 is temporarily stored in the self, and as shown in FIG. 3, it is scanned pixel by pixel in the vertical direction and output to the subsequent stage. Such data scanning is called belt scanning.

  The Bayer interpolation block 5 performs interpolation processing from the Bayer data (see FIG. 4A) transferred from the first DMA 4, and finally outputs it as YUV data. In this embodiment, as shown in FIG. 4B, YUV data of one pixel is generated from Bayer data of 5 × 5 pixels.

  The enlargement block 8 enlarges the YUV data input via the switch 7 to a desired magnification (increases the number of pixels) in accordance with an instruction from the control unit 2 and is connected in series in a parallel state on the rear stage side. Selectively output to the first reduced block 9 or the second reduced block 10. Both the input and output of data in the enlargement block 8 is belt scanning. The magnification can be set independently for horizontal and vertical, and the enlargement algorithm is linear interpolation that obtains an interpolated result of one pixel from six neighboring pixels in both the horizontal and vertical directions. FIG. 5 shows an example of vertical 5/3 times. For example, when the interpolation point is between the pixel “2” and the pixel “3” in the original image (“2.3” or “2.9” in the figure), the pixel “0-5” is used as the interpolation coefficient Lanczos. Is used to find the interpolation point (output pixel). The setting of the magnification is designated by the pixel interval when one pixel in the original image is divided into 16384. For example, when it is 2 times, it is 8192, when it is 4 times. Therefore, an arbitrary magnification setting that can be expressed with this resolution is possible.

  The first reduction block 9 and the second reduction block 10 reduce the input YUV data and output it. The reduction algorithm is an average operation method, and both data input and output is belt scanning. The magnification can be set independently for both horizontal and vertical. When the magnification is n / m (where n ≦ m), n can be set within the range of 1 to 256 and m is 1 to 256 for the horizontal magnification, and n is 1 to 16 and m is 1 to 1 for the vertical magnification. It can be set within the range of 32. The vertical magnifications m and n are the number of lines on the input side and the number of lines on the output side of the belt to be processed, respectively. FIG. 6 shows an example in which 6 vertical lines are reduced to 4 lines.

  The first reduced block 9 and the second reduced block 10 are completely equivalent blocks, and the magnification setting is such that both the reduced blocks 9 and 10 share the same denominator (m). Different magnifications can be set. However, the numerator of the magnification of the second reduction block 10 (= the number of lines of the output belt) is “8” or “16” depending on the JPEG encoding format because of the connection with the JPEG input I / F 12 at the subsequent stage. Will be selected from.

  The third DMA 11 is a data transfer unit that receives image data by belt scanning from the first reduction block 9 in accordance with an instruction from the control unit 2 and transfers the image data to a corresponding portion on the memory 3.

  The JPEG input I / F 12 temporarily holds image data (YUV data) input by the belt scan in pixel block units (MCU), and conforms to the input format (Y: Cb: Cr) on the JPEG encoder 13 side. In this format, the interface unit converts the data into 4: 2: 2, 4: 2: 0, etc. and outputs the data. That is, when the input format is 4: 2: 0, 16 × 16 pixels are output as a pixel block unit as shown in FIG. 7, and when the input format is 4: 2: 2, Outputs 8 pixels as a pixel block unit. The JPEG encoder 13 JPEG-encodes the input YUV data in a 4: 2: 2 or 4: 2: 0 format, and outputs code data. The fourth DMA 14 is a data transfer unit that transfers the code data to a corresponding part on the memory 3.

  Next, processing contents of the control unit 2 will be described. First, processing contents relating to data transfer under the control of the first DMA 4 and the second DMA 6 will be described. FIG. 8 is a flowchart showing the procedure. The control unit 2 initializes the belt number i to be processed at the start of processing (step SA1), and then corresponds to the left end, right end, upper end, and lower end of the i-th belt (0th at the beginning of processing). The coordinates (ys_left, ys_rgt, ys_top, ys_btm) on the original image (Bayer data or YUV data) are calculated (step SA2). In the present embodiment, the following calculation is performed.

Here, as an example, it is assumed that enlargement processing is performed on YUV data when the digital zoom function is set. For convenience, the enlargement magnification is 5 times both horizontally and vertically, and the size of the enlarged image is 32 × 32 pixels. The description will be made assuming that the number of belt lines is 8 (= the configuration in which the number of belts in the enlarged image is 4). FIG. 9 shows the relationship between the original image (YUV data) at the time of belt processing on the premise and the belt (0 to 3) in the corresponding enlarged image (image finally input to the JPEG encoder) C. FIG. In the figure, the large circles are pixels of the original image, and the lightly-painted circles are the processing belts, that is, the pixels in the range of the 0th belt in FIG. 5A and the 1st belt in FIG. . A small black circle is a pixel in the output belt range (32 × 8 pixels) with respect to the processing belt.
The coordinates of each pixel in the enlarged image C are (xb, yb),
The reciprocal of the magnification in the horizontal and vertical directions (xstep, ystep),
The position on the original image corresponding to the upper left corner on the enlarged image C is (x0, y0)
When
The coordinates corresponding to the original image (xs, ys) are
xs = xb × xstep + x0
ys = yb × ystep + y0
Can be obtained.

  In the present embodiment, the coordinate positions of the left end, the right end, the upper end, and the lower end of each belt are calculated according to the above formulas. The calculation result is rounded down to the final calculation result.

  FIG. 10 shows the coordinates of the left end, right end, upper end, and lower end of the enlarged image C in each belt shown in FIG. 9 obtained by the above calculation, and the left end, right end, upper end, and lower end of the original image obtained directly from the above formula. It is the figure which showed the correspondence with the coordinate of a coordinate and the left end, right end, upper end, and lower end as an input range used as the final calculation result. Of the values of each coordinate shown in the figure, the position (x0, y0) on the original image corresponding to the upper left end on the enlarged image C, which is necessary for calculating the original image corresponding coordinates (xs, ys), that is, The upper end (vertical coordinate: “0.8”) and the left end (horizontal coordinate: “1.5”) of the coordinates corresponding to the original image in the belt 0 are coordinate values determined by the relationship with the enlargement ratio, and this embodiment Is a coordinate value obtained from a correspondence table prepared in advance in the program ROM of the control unit 2 or the like.

  Subsequently, coordinates obtained by expanding the peripheral pixels necessary for the enlargement processing are obtained with respect to the coordinates obtained as described above (step SA3). Here, it is assumed that each pixel is expanded by three pixels vertically and horizontally. Next, when the original image is Bayer data and Bayer interpolation is required (YES in Step SA4), coordinates obtained by expanding peripheral pixels necessary for Bayer interpolation processing in the Bayer interpolation block 5 are obtained (Step SA5). Here, it is assumed that two pixels are expanded vertically and horizontally. Then, the obtained coordinates are set in the first DMA 4 as an input area of the original image, and belt processing (data transfer by belt scanning) is started (step SA6). If the original image is not Bayer data but YUV data (NO in step SA4), the coordinates obtained in step S3 are set in the second DMA 6 as the input area of the original image, and then belt processing is started (step SA7). . When the belt process is completed (YES in step SA8), the belt number is updated (incremented) (step SA9). Thereafter, returning to step SA2, the above-described processing is repeated, and when the processing for the belt with the last belt number, that is, processing for one image is completed (step SA10), the processing is terminated.

  FIG. 11 is a transition diagram showing changes in the data content of each belt in each unit of the image processing apparatus 1 due to the processing of the control unit 2 described above. That is, the interpolation input belt Bayer data a in the original image A transferred from the memory 3 is subjected to pixel interpolation in the Bayer interpolation block 5 and then transferred to the enlargement block 8 as the enlargement input belt YUV data b. Further, the enlarged output belt (base belt) YUV data c subjected to enlargement (increasing the number of pixels) processing is transferred to the first reduction block 9 and the second reduction block 10 and reduced at different reduction ratios. The first output belt YUV data d and the second output belt YUV data e are output, and the second output belt YUV data e is input to the JPEG encoder 13 via the JPEG input I / F 12.

  Then, in the above-described belt processing (data transfer), the left end, the right end, the upper end, and the lower end of each belt (in FIG. 10, the coordinates corresponding to the original image on the vertical side) as shown in FIG. ) Are individually determined, that is, the number of lines is individually controlled, whereby the second output belt YUV data e input from the second reduction block 10 to the JPEG encoder 13 via the JPEG input I / F 12 The number of lines is maintained at the number of lines (eight lines here) according to the input format on the JPEG encoder 13 side. Of course, the above calculation is based on the enlargement factor, and the enlargement factor is reflected in the number of lines to be controlled. Therefore, any enlargement factor (or reduction factor) is set, not limited to the aforementioned enlargement factor. Even in such a case, the number of lines of the second output belt YUV data e can be set to the number of lines according to the input format on the JPEG encoder 13 side.

  Next, processing contents when the control unit 2 sets the magnification for the above-described enlarged block 8, first reduced block 9, and second reduced block 10 will be described. FIG. 12 is a flowchart showing the procedure. The magnification to be set corresponds to the size of the original image (Bayer data or YUV data), for example, according to the magnification required by the digital zoom function at that time. Here, only the setting of the magnification in the vertical direction will be described.

  As the process starts, the control unit 2 determines the height of the belt determined by the JPEG encoding format at that time, that is, the number of lines (“16” or “8”), and the requested JPEG input image (JPEG encoder 13 The number of belts is calculated from the size of the image input to (step SB1). Next, the average belt height of the original image is calculated from the calculated number of belts and the height of the original image (step SB2), and it is determined whether or not it is larger than the belt height of the JPEG input image (step SB3). ).

  If the result of this determination is NO and the average belt height of the original image is equal to or less than the belt height of the JPEG input image, it is determined that only enlargement processing is necessary, and the belt height of the enlarged image is determined based on the height of the JPEG input image. As the belt height (step SB4), the vertical magnification of the enlargement block 8 is set to the magnification obtained from the ratio between the belt height of the JPEG input image and the average belt height of the original image (step SB5), and the second reduction. The vertical magnification of the block 10 is set to “1 ×” (step SB6).

  When the determination result in step SB3 is YES and the average belt height of the original image exceeds the belt height of the JPEG input image, the calculation result of the average belt height of the original image in step SB2 is further divisible. It is determined whether or not the value is the same (step SB7). Note that the function int (r) used at this time is a function for truncating the decimal part. If it is divisible (YES in step SB7), it is determined that only the reduction process is necessary, and the belt height of the enlarged image is left as it is (average of belt height of the original image) (step SB8). The magnification is set to “1 ×” (step SB9), and the vertical magnification of the second reduction block 10 is set to a magnification that makes the belt height of the enlarged image the belt height of the JPEG input image (step SB10).

  If the determination result in step SB7 is NO and the calculation result of the average belt height of the original image in step SB2 is a value that cannot be divided, it is determined that the image needs to be enlarged and reduced. The belt height is changed to a rounded up value (step SB11). Note that the function roundup (r) used at this time is a function that rounds up after the decimal point. Next, the calculated height of the enlarged image is calculated from the belt height and the number of belts of the enlarged image after the change (step SB12). Then, the vertical magnification of the enlarged block 8 is set to a magnification obtained from the ratio between the height of the enlarged image calculated and the height of the original image (step SB13), and the vertical magnification of the second reduced block 10 is set to the enlarged image. Is set to a magnification that sets the belt height of the JPEG input image as the belt height (step SB14).

  In this way, a combination of magnifications corresponding to the magnifications requested at that time is set for each of the enlarged block 8 and the second reduced block 10, and finally output from the second reduced block 10. The number of belts and the belt height of the enlarged image are values corresponding to the required magnification. That is, the size of the enlarged image becomes the size of the processing target corresponding to the requested magnification.

  When any of the above-described steps SB6, SB10, and SB14 is completed, the preview reduction setting process shown in FIG. 13 is performed (step SB15). In such processing, first, the average belt height of the preview original image is calculated from the number of belts of the JPEG input image acquired in step SB1 and the height of the determined preview image (display YUV data) (step SB101). . It is determined whether or not the calculation result is a divisible value (step SB102). If it is divisible (YES in step SB102), the vertical magnification of the first reduction block 9 is set to a magnification that makes the belt height of the preview original image the belt height of the JPEG input image (step SB104). If it is not divisible (NO in step SB102), the average belt height of the preview original image is rounded up and changed to a value obtained by rounding up the belt height of the preview original image (step SB103), and then input to JPEG The magnification is set as the belt height of the image (step SB104). Thereby, the control part 2 complete | finishes magnification setting.

  In FIG. 14, it is determined that it is necessary to enlarge and reduce in the above magnification setting process, and when the processes of steps SB11 to SB14 are performed, the enlarged block 8, the first reduced block 9, and the second reduced block 10 are processed. It is the figure which showed the data content of the combination of a specific setting magnification, the original image a at that time, the enlarged image b, the JPEG input image c, and the preview original image d. Here, the JPEG encoding format is 4: 2: 0, the size of the original image a is 3 million pixels (2048 × 1536), and the size of the JPEG input image c corresponding to the requested magnification (reduction) is 2 million pixels ( 1600 × 1200) and the preview original image d is QVGA (240 × 320).

  In such a case, the horizontal magnification of the enlargement block 8 is the same magnification, the vertical magnification is 1.025 ..times, the size of the enlarged image b is 2048 × 1575, the number of belts is 75, and the belt height is 21. . The second reduction block 10 has a horizontal magnification of 100/128 times and a vertical magnification of 16/21 times. The size of the JPEG input image c is 1600 × 1200, the number of belts is 75, the belt height is 16 At the same time, the horizontal magnification of the first reduction block 9 is 20/128 times, the vertical magnification is 4/21 times, the size of the preview original image d is 320 × 300 which is slightly larger than QVGA, the number of belts is 75, the belt height Is 4.

  Next, a still image encoding process performed by the control unit 2 based on the above-described data transfer process and magnification setting process at the time of still image shooting with a digital camera will be described. FIG. 15 is a flowchart showing the procedure. That is, the control unit 2 generates display YUV data (preview image) from the Bayer data by the system using the first reduction block 9 by the first data conversion by the data conversion based on the series of data transfer described above. At the same time, the main image is JPEG-encoded from the Bayer data by the system using the second reduced block 10, and the encoded data is output to the memory 3 (step SC1). In the second data conversion, JPEG encoding of the thumbnail image associated with the main image is performed from the display YUV data stored in the memory 3 by the first conversion by the system using the second reduced block 10. Then, the encoded data is output to the memory 3 (step SC2).

  In the image processing apparatus 1 described above, when the image data stored in the memory 3 is enlarged or reduced for each belt and JPEG encoding is performed as described above, any enlargement ratio (or reduction ratio) is set. Even if it is, the number of lines of the belt (second output belt YUV data e) input to the JPEG encoder 13 can be set to the number of lines set in the JPEG encoder 13. Therefore, it is necessary to provide a circuit such as a buffer memory for ensuring the consistency of the number of lines in the second reduced block 10 even if the JPEG encoder 13 exists on the subsequent stage side of the second reduced block 10. In other words, it is possible to perform zooming processing with various magnification settings at high speed with a simple circuit configuration.

  Such an effect can also be obtained in a configuration in which a single scaling block 52 is provided as a processing block for enlarging or reducing an image, such as the image processing apparatus 51 shown in FIG. it can. In the present embodiment, the first DMA 4 and the second DMA 6 perform belt scanning for each pixel. However, the belt scanning may be performed with two or more pixels. Further, although the image data is transferred separately to the Bayer interpolation block 5 and the switch 7 (enlarged block 8) using the two first DMAs 4 and the second DMA 6, they may be combined into one. In addition, although only the Bayer interpolation block 5 exists in the preceding stage of the enlarged block 8, a plurality of processing blocks may be provided in series in the preceding stage of the enlarged block 8, or may be provided in parallel and switched. It may be configured to be selectively connected by a switch or the like.

  In the present embodiment, as described above, the original image is reduced and / or reduced at the required magnification by performing the continuous enlargement process and the reduction process in both the enlargement block 8 and the second reduction block 10. It is the structure which expands. Therefore, if the requested magnification is a reduction ratio that causes image degradation in the second reduction block 10, the magnification assigned to the enlargement block 8 and the second reduction block 10 is set to such reduction. By adjusting the rate so as not to set the second reduced block 10, it is possible to perform a reduction process that does not cause image degradation. Therefore, a high-quality reduced image can always be obtained. In addition, since the enlargement process in the enlargement block 8 and the reduction process in the second reduction block 10 are performed using algorithms suitable for each process, a higher-quality enlarged image or reduced image can be obtained. Although the interpolation coefficient used in the enlarged block 8 is Lanczos, other interpolation coefficients can be used.

  Moreover, in the present embodiment, since the first reduced block 9 and the second reduced block 10 are connected in parallel to the enlarged block 8, images of different sizes from the same original image can be efficiently processed in a short time. Can be generated. As a result, at the time of still image shooting, as described above, the main image, the accompanying thumbnail image, and the preview image can be generated by two data conversions, and accordingly, the shooting interval in the digital camera can be shortened. Can do. In the present embodiment, the main image and the preview image are generated by the first data conversion. However, other images having different purposes may be generated. For example, an image for code amount measurement may be generated instead of a thumbnail image and encoded by the JPEG encoder 13, and in this case, the main image is generated by the second data conversion. Also good. Furthermore, although not shown, in addition to the first reduced block 9 and the second reduced block 10, other reduced blocks may be provided in parallel therewith. In such a case, the above-described two data conversion processes can be completed with one data conversion.

  In the present embodiment, the enlarged or reduced image data is YUV data, but the enlarged or reduced image data may be RGB image data. Further, although the image data after enlargement or reduction is JPEG encoded, it may be compressed and encoded by another method such as MPEG, and the JPEG encoder 13 may have the number of input lines. It may be another processing block in which is restricted.

  Furthermore, in the present embodiment, the image processing apparatus 1 incorporated in a digital camera has been described. However, the present invention is incorporated in other video equipment, an image recognition apparatus, or the like, or used alone. It can also be applied to things.

1 is a circuit diagram showing a schematic configuration of an image processing apparatus according to the present invention. It is a schematic diagram which shows the belt used as an image processing unit. It is a figure which shows the transfer method of the data read from memory. (A) is a figure which shows the example of Bayer data, (b) is a conceptual diagram when producing | generating YUV data from Bayer data. It is a figure which shows the specific example of the correspondence of the pixel of an original image in an expansion process, and an interpolation point. It is a figure which shows the specific example of the number of lines of the input side belt and output side belt in a reduction process. It is a schematic diagram which shows the input / output data in a JPEG encoder when a JPEG encoding format is 4: 2: 0. It is a flowchart which shows the processing content regarding the data transfer by a control part. It is explanatory drawing which shows the specific example when setting the input range of each belt with respect to an original image. It is the figure which showed the relationship between the coordinate position of the enlarged image in each belt of FIG. 9, and the coordinate position of an original image. It is the transition figure which showed the change of the data content of each belt in each part of the image processing device. It is a flowchart which shows the processing content regarding the magnification setting by a control part. It is a flowchart which shows the reduction setting process for a preview by a control part. It is a figure which shows the specific example of the image data produced | generated with the magnification set by a magnification setting process and data transfer. It is a flowchart which shows the still image encoding process by a control part. It is a circuit diagram which shows schematic structure of another image processing apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Control part 3 Memory 4 1st DMA
5 Bayer interpolation block 6 2nd DMA
7 Switch 8 Enlarged block 9 First reduced block 10 Second reduced block 11 Third DMA
12 JPEG input I / F
13 JPEG encoder 14 4th DMA
A All images B Image data (belt)

Claims (8)

An enlargement processing unit for enlarging the image data of the transferred original image;
A reduction processing unit that is connected in series to the subsequent stage of the enlargement processing unit, and that reduces the image data after the enlargement processing by the enlargement processing unit,
An image processing apparatus comprising: a magnification setting unit configured to set a combination magnification according to a requested magnification in the enlargement processing unit and the reduction processing unit.   The image processing apparatus according to claim 1, wherein the reduction processing unit reduces the image data after the enlargement processing by the enlargement processing unit based on an image processing algorithm different from the enlargement processing.   3. The reduction processing unit according to claim 1, wherein a plurality of the reduction processing units are connected in parallel at a subsequent stage of the enlargement processing unit, and the magnification setting unit sets different reduction ratios to the plurality of reduction processing units, respectively. Image processing device.   The image processing apparatus according to claim 1, wherein the reduction processing unit outputs the image data after the reduction process to a compression unit that compresses and encodes the image data. Transfer means for transferring the image data of the original image to the enlargement processing unit in units of processing consisting of a plurality of lines;
Line number setting means for setting the number of output lines of image data in the reduction processing unit;
An image processing apparatus comprising: a line number control unit configured to control the number of transfer lines of image data transferred to the enlargement processing unit by the transfer unit for each processing unit based on the requested magnification. .   The line number control means includes the requested reduction ratio, an upper end position of an original image area corresponding to the requested reduction ratio to which the transfer means should transfer image data in an image based on the image data, Based on the coordinate positions of the upper end and the lower end of each processing unit corresponding to the number of output lines in the image having a reduced size with the requested reduction rate, the image data transferred to the enlargement processing unit by the transfer unit 6. The image processing apparatus according to claim 5, wherein the number of transfer lines is controlled. In the image processing method for scaling the image data at the requested magnification,
Setting a combination of an enlargement ratio and a reduction ratio according to the requested magnification;
A process of enlarging the image data at the set enlargement rate;
And a step of reducing the image data after the enlargement processing at the set reduction rate. An enlargement processing unit for enlarging the transferred image data of the original image, and an image processing that is connected in series to the subsequent stage of the enlargement processing unit, and the image data after the enlargement processing by the enlargement processing unit is different from the enlargement processing A computer having an image processing apparatus including a reduction processing unit that performs reduction processing based on an algorithm functions as a magnification setting unit that sets a combination of magnifications according to a requested magnification in the enlargement processing unit and the reduction processing unit Program to let you.

Images