JP2016105329A - Data processing apparatus and data processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data processing apparatus configured to reduce power consumption for data transfer, while maintaining data transfer efficiency, and a data processing method.SOLUTION: A data processing apparatus includes: a first data generation section which sets a first bus area and a second bus area on a data bus to divide packed data into a first pack area and a second pack area, and extracts first data to be packed in the first pack area without space, so as to be transferred from the same position in the first bus area during burst transfer of the first data; a second data generation section which packs second data, which is input data left by the first data generation section, in the second pack area; and a data connection section which connects the first data packed in the first pack area with the second data packed in the second pack area to generate packed data.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a data processing apparatus and a data processing method.

  In an imaging device such as a still image camera, a video camera, a medical endoscope camera, or an industrial endoscope camera, as the number of pixels and the speed of the imaging device improve, Image data including “pixel data”). In such an imaging apparatus, a memory for temporarily storing data is used when each processing block in the imaging apparatus processes image data obtained by imaging. The memory temporarily stores image data at each processing stage.

FIG. 14 is a block diagram illustrating a schematic configuration of a conventional imaging apparatus. For example, in the shooting operation in the imaging apparatus shown in FIG. 14, image data processing is performed in the following procedure.
(Procedure 1): First, the imaging processing unit, for example, stores image data obtained by a CCD (Charge Coupled Device) type solid-state imaging device in a memory via an output DMA (Direct Memory Access) unit. Transfer and store temporarily.
(Procedure 2): Subsequently, the image processing unit reads out the image data temporarily stored in the memory via the input DMA unit. The image processing unit performs image processing for recording and image processing for display on the read image data. Thereafter, the image data after the image processing is transferred to the memory via the output DMA unit and temporarily stored.
(Procedure 3): Subsequently, the display processing unit reads out the image data subjected to the image processing for display via the input DMA unit, and displays it on the display device.

  As described above, in the imaging apparatus, the preceding processing block temporarily stores the image data in the memory. Then, the subsequent processing block reads the image data recorded in the memory and performs the next processing. In this manner, each processing block in the imaging apparatus sequentially performs the entire processing of the imaging apparatus by transferring the image data to be processed via the memory.

  In recent years, it has been desired that imaging devices such as still image cameras and moving image cameras can be used continuously for a long time, and a technique for reducing the power consumption of an electric circuit in the imaging device is required. One method for reducing the power consumption of the imaging apparatus is a method of increasing the transfer speed of image data between each processing block (electric circuit) and the memory. Increasing the transfer rate of image data can be realized, for example, by increasing the frequency of the clock at which the imaging apparatus operates, or by shortening the transfer period of image data between each processing block and the memory. These methods reduce the power consumption for transferring image data by increasing the transfer rate of image data.

  As a technique for shortening the transfer period of image data between a processing block and a memory, a technique called packing as disclosed in Patent Document 1 is known. The packing technique disclosed in Patent Document 1 expands the bus width of a data bus used when transferring each pixel data in image data to a memory, and arranges adjacent pixel data on the data bus (packing). ) To transfer a plurality of pixel data at a time. With this technique, the number of times of data transfer required to transfer all pixel data can be reduced compared to the data transfer before transferring pixel data for each pixel, and the period related to data transfer of image data Can be shortened. For example, when pixel data obtained from a 16-row × 16-column Bayer array CCD as shown in FIG. 15 is transferred to a memory, the packing technique disclosed in Patent Document 1 uses pixel data for four pixels. By using as a unit for transferring at a time, the transfer period of image data can be shortened to ¼. Thereby, the power consumption of the electric circuit in the imaging device can be reduced as compared with the case where the pixel data is transferred to the memory pixel by pixel.

  Further, as a technique for further shortening the transfer period of image data, a packing method using DMA burst transfer is conceivable. In this method, one burst, which is a unit for accessing the memory at a predetermined number of cycles, is considered as a unit for packing pixel data. FIG. 16 shows an example of packing of pixel data. FIG. 16A shows an example in which the pixel data shown in FIG. 15 is packed by the packing method disclosed in Patent Document 1, and FIG. 16B shows the pixels shown in FIG. 15 in units of bursts. An example of packing data is shown. In FIG. 16, a bus width (hereinafter referred to as “memory bus width”) of a data bus (hereinafter referred to as “memory bus width”) used for transferring pixel data to the memory is set to 32 bits, and one cycle in burst transfer. In this example, memory access for one minute (hereinafter referred to as “one transfer”) is performed four times, that is, one burst transfer is performed with four transfers. In FIGS. 16A and 16B, an example in which the resolution of pixel data of one pixel, that is, the number of bits of pixel data is 9 bits, 10 bits, 12 bits, and 14 bits, in order from the top, respectively. Is shown.

  As can be seen from FIG. 16, in the packing method disclosed in Patent Document 1 shown in FIG. 16A, pixel data for two pixels per transfer is arranged on the memory bus, and pixels for eight pixels per burst. Data can be transferred to memory. On the other hand, in the packing method in units of bursts shown in FIG. 16B, pixel data for 14 pixels, 12 pixels, 10 pixels, and 9 pixels can be transferred to the memory. This is because the packing method disclosed in Patent Document 1 arranges (packes) pixel data in units of one memory transfer, that is, the number of bits of a plurality of arranged pixel data. It is necessary not to exceed. For this reason, in the packing method disclosed in Patent Document 1, there are many bits (hereinafter referred to as “unused bits”) in which pixel data cannot be allocated within the memory bus width. On the other hand, in the packing method in units of bursts, pixel data is arranged (packed) in units of one burst. Therefore, as shown in FIG. 16B, the total number of bits of a plurality of pixel data to be arranged is the memory bus. Even when the width is exceeded, the pixel data can be arranged (packed) for the next transfer if one burst is not exceeded. That is, in the packing method in units of bursts, even when the memory bus width is not an integral multiple of the resolution of pixel data, pixel data can be arranged (mapped) across one transfer unit. Can be reduced. Thus, the packing method in units of bursts can transfer more pixel data to the memory in the same time than the packing method disclosed in Patent Document 1, that is, the image data transfer period can be shortened. The power consumption of the electric circuit in the imaging device can be further reduced.

JP 2007-31358 A

  Generally, it is known that the power consumption is smaller when the data change (data “0” → “1” or “1” → “0” change (inversion)) is smaller. For this reason, it is conceivable to reduce the power consumption of the imaging apparatus by reducing the change in data on the memory bus between each processing block (electric circuit) in the imaging apparatus and the memory. FIG. 17 is a diagram for explaining the relationship between the change in data on the data bus (memory bus) between the processing block in the imaging apparatus and the memory and the power consumption. FIG. 17A illustrates an example in which the bus width of the memory bus between the imaging processing unit and the memory in the imaging apparatus illustrated in FIG. 14 is 32 bits. 17B-1 to 17B-3 schematically show changes in data on the memory bus. In the example of FIG. 17, the case of FIG. 17 (b-1) where there is no change in data on the memory bus has the least power consumption and the change in data on the memory bus is the largest. In this case, the power consumption is the largest.

  Therefore, in burst transfer between each processing block in the imaging apparatus and the memory, pixel data between two successive transfers (for example, pixel data of the first transfer and the second transfer of each burst transfer shown in FIG. 16). It can be seen that the power consumption associated with the transfer of the image data also increases when there is a large change in (). That is, the power consumption related to the transfer of image data changes in proportion to the number (number of bits) in which the same bit on the memory bus changes between two transfers.

  In general, image data has a small amount of data change between adjacent pixels, and when compared with the same bit of each pixel data of adjacent pixels, the value of the inverted bit is inverted. It is expected that there will be fewer than not bits. Here, comparing the packing method disclosed in Patent Document 1 shown in FIG. 16 and the packing method in units of bursts, as shown in FIG. 18, the same bits are arranged in the same color pixel data. The packing method disclosed in Patent Document 1 is considered to consume less power for transferring image data. FIG. 18 shows a case where the number of bits of pixel data of one pixel is 9 bits in the packing method disclosed in Patent Document 1 shown in FIG. 16 and the packing method in units of bursts.

  More specifically, in the packing method disclosed in Patent Document 1 shown in FIG. 18A, the least significant bits of the memory bus shown in the range A are all the same bits (the least significant bits) of pixel data of the same color. Bit). On the other hand, in the packing method in units of bursts shown in FIG. 18B, the least significant bits of the memory bus shown in range B are all different bits of pixel data of different colors. Therefore, by packing image data in units of bursts, if the position of the bit of pixel data arranged on the memory bus differs between two consecutive transfers, the amount of change of the same bit on the memory bus This increases the power consumption associated with the transfer of image data.

  That is, the packing method disclosed in Patent Document 1 shown in FIG. 18A consumes less power for transferring image data, but has low data transfer efficiency. The packing method in units of bursts shown in FIG. 18B has good data transfer efficiency, but consumes a lot of power for transferring image data.

  As described above, in the packing of pixel data in units of bursts, by placing a larger amount of pixel data on the memory bus, the pixel data transfer period is shortened, resulting in a reduction in power consumption for image data transfer. However, since the positions of the bits of the pixel data arranged on the memory bus are different between two successive transfers, the effect of reducing the power consumption related to the transfer of image data can be sufficiently obtained. There is no problem.

  In the packing method disclosed in Patent Document 1, for example, as shown in FIG. 19, pixel data for three pixels can be arranged (packed) in one transfer for packing pixel data. However, in this case, for example, the least significant bit of the memory bus shown in the range C is the same bit (the least significant bit) of the pixel data, but becomes pixel data of a different color. The pixel data is likely to have a large difference in value when the color is different. Even with the packing method disclosed in Patent Document 1, the power consumption associated with the transfer of the image data is reduced depending on the arrangement of the pixel data in one transfer. It will not be done.

  The present invention has been made on the basis of the above-mentioned problem recognition, and provides a data processing device and a data processing method capable of reducing power consumption related to data transfer while maintaining data transfer efficiency. It is an object.

  In order to solve the above problems, a data processing apparatus of the present invention is a data processing apparatus that generates pack data for burst transfer of a plurality of input data on a data bus, and the data processing apparatus includes a first data processor on the data bus. By setting the bus area and the second bus area, the pack data is divided into the first pack area and the second pack area, the first data is taken out, and the burst of the first data is obtained. A first data generation unit that packs the first pack area without a gap so that it is transferred from the same position in the first bus area at the time of transfer, and the input data left by the first data generation part. A second data generation unit that packs certain second data into the second pack area; and the first data packed into the first pack area and the second pack area. And a data combination unit that generates the pack data by combining the second data that has been processed, and the plurality of input data is image data having a plurality of color information, In the data between two successive transfers in the first pack area, adjacent pixel data of the same color are arranged in the same bit on the memory bus.

  Further, the second data generation unit of the present invention is characterized in that the second data is packed so as to be continuous in the second pack area.

  The data processing device of the present invention is a data processing device for restoring packed original data from packed data burst-transferred on a data bus having a predetermined number of bits. A bus area and a second bus area are set, the data of the first pack area of the pack data is extracted from the first bus area and output, and the second of the pack data is output from the second bus area. A data dividing unit for extracting and outputting data of the pack area, a first data restoring unit for restoring the first data based on the data extracted from the first pack area, and the second pack area A second data restoring unit for restoring the second data based on the data extracted from the data, and outputting the restored first data and the original data based on the second data Data selection unit, and the first data is data packed in the first pack area without a gap so as to be transferred from the same position in the first bus area during the burst transfer. The original data is image data having a plurality of pieces of color information, and in the data between two successive transfers of the first pack area in the pack data, adjacent pixel data of the same color are stored on the memory bus. Are arranged in the same bit.

  Further, the second data of the present invention is data packed so as to be continuous in the second packed area.

  The data processing method of the present invention is a data processing method for generating packed data for burst transfer of a plurality of input data onto a data bus, wherein the first bus area and the second data are generated on the data bus. And setting the bus area, the step of dividing the pack data into the first pack area and the second pack area, the first data is taken out, and the first data is burst transferred during the burst transfer of the first data. Packing the first pack area without gaps so as to be transferred from the same position in one bus area, and the second data as the input data remaining from the first data in the second pack A step of packing in a region, and combining the first data packed in the first packed region and the second data packed in the second packed region Generating the pack data, and the plurality of input data is image data having a plurality of color information, and data between two successive transfers of the first pack area in the pack data 2 is characterized in that adjacent pixel data of the same color are data arranged in the same bit on the memory bus.

  The data processing method of the present invention is a data processing method for restoring packed original data from packed data burst-transferred on a data bus having a predetermined number of bits. A step of setting a bus area and a second bus area; a step of extracting and outputting data of the first pack area of the pack data from the first bus area; and the pack from the second bus area Extracting and outputting data of the second packed area of data; restoring the first data based on the data extracted from the first packed area; and extracting from the second packed area Restoring the second data based on the recovered data, and outputting the restored first data and the original data based on the second data And the first data is data packed in the first pack area without a gap so as to be transferred from the same position in the first bus area during the burst transfer, The original data is image data having a plurality of color information, and in the data between two successive transfers in the first pack area in the pack data, adjacent pixel data of the same color are the same bit on the memory bus. It is the data arranged in.

  According to the present invention, it is possible to reduce power consumption for data transfer while maintaining data transfer efficiency.

1 is a block diagram illustrating a schematic configuration of an imaging apparatus according to an embodiment of the present invention. It is the block diagram which showed schematic structure of the data conversion part of the 1st structure with which the imaging device of this embodiment was equipped. It is the block diagram which showed schematic structure of the data reverse conversion part of the 1st structure with which the imaging device of this embodiment was equipped. It is a figure explaining the 1st data arrangement method in the data converter of the 1st composition. It is a figure explaining the 2nd data arrangement method in the data conversion part of the 1st composition. It is a figure explaining the 3rd data arrangement method in the data conversion part of the 1st composition. It is the block diagram which showed schematic structure of the data conversion part of the 2nd structure with which the imaging device of this embodiment was equipped. It is the block diagram which showed schematic structure of the data reverse conversion part of the 2nd structure with which the imaging device of this embodiment was equipped. It is a figure explaining the 4th data arrangement method in the data conversion part of this 2nd composition. It is the figure which showed an example at the time of applying the data arrangement method by the data converter with which the imaging device of this embodiment was equipped to another image data. It is the figure which showed an example at the time of applying the data arrangement method by the data conversion part with which the imaging device of this embodiment was equipped to another image data. It is the figure which showed an example at the time of applying the data arrangement method by the data conversion part with which the imaging device of this embodiment was equipped to another image data. It is the figure which showed an example at the time of applying the data arrangement method by the data conversion part with which the imaging device of this embodiment was equipped to another data. It is the block diagram which showed schematic structure of the conventional imaging device. It is the figure which showed an example of the arrangement | sequence of image data. It is the figure which showed an example of the conventional data packing. It is a figure explaining the relationship between the change of the data on a data bus, and power consumption. It is the figure which showed an example of the state of the data arrangement | positioning in the conventional data packing method. It is the figure which showed another example of the state of data arrangement in the conventional data packing method.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a schematic configuration of an imaging apparatus according to the present embodiment. 1 includes a CPU 10, a memory 20, a CCD 30, an imaging processing unit 40, an image processing unit 50, a display processing unit 60, a display device 70, and data conversion units 41 and 51. , Data inverse conversion units 52 and 62, output DMA units 45 and 55, and input DMA units 56 and 66. Note that the imaging apparatus 1 also has a function of recording captured image data, but in the following description, a case where captured image data is displayed will be described for ease of explanation. Further, FIG. 1 shows only the connection of a memory bus that is a data bus related to the transfer of pixel data between each component (processing block) in the imaging apparatus 1.

The CPU 10 is a control device that performs overall control of the imaging device 1.
The memory 20 is a memory such as a DRAM (Dynamic Random Access Memory) that temporarily stores pixel data processed by each processing block in the imaging apparatus 1. The memory 20 temporarily stores pixel data of each processing stage by each processing block in the imaging apparatus 1.

The CCD 30 is, for example, a Bayer array solid-state imaging device that converts incident subject light into an image signal. The CCD 30 outputs the pixel signal of the subject to the imaging processing unit 40.
The imaging processing unit 40 generates image data obtained by performing predetermined signal processing on the pixel signal input from the CCD 30.

The image processing unit 50 generates image data obtained by performing various image processes in the imaging device 1 on the image data generated by the imaging processing unit 40.
The display processing unit 60 converts the image data image-processed by the image processing unit 50 into display data corresponding to the display device 70. The display processing unit 60 outputs the converted display data to the display device 70.
The display device 70 is a display device that displays display data, such as a liquid crystal display.

  In the imaging apparatus 1 of this embodiment, the pixel signal of the subject imaged by the CCD 30 is sequentially processed by the imaging processing unit 40, the image processing unit 50, and the display processing unit 60 and displayed on the display device 70. At this time, the image data at each processing stage is transferred between the processing blocks via the memory 20. The data conversion units 41 and 51, the data reverse conversion units 52 and 62, the output DMA units 45 and 55, and the input DMA units 56 and 66 in the imaging apparatus 1 are processes related to the transfer of image data at each processing stage. It is a block.

  The data conversion units 41 and 51 have predetermined pixel data in the image data input from the preceding processing block (in the present embodiment shown in FIG. 1, the imaging processing unit 40 or the image processing unit 50). It is arranged at each bit of the memory bus by the arrangement (packing) method. A detailed description of the configuration of the data conversion units 41 and 51 and the pixel data packing method by the data conversion units 41 and 51 will be described later.

The output DMA units 45 and 55 write (store) image data composed of pixel data packed into each bit of the memory bus by the data conversion unit 41 or 51 in the memory 20 by DMA access.
The input DMA units 56 and 66 read the image data stored in the memory 20 by DMA access, and output the read image data to the data reverse conversion unit 52 or 62.

  The data inverse conversion units 52 and 62 return the pixel data included in the image data input from the input DMA unit 56 or 66 to the original image data by a method reverse to a predetermined arrangement (packing) method. To the processing block (in the present embodiment shown in FIG. 1, the image processing unit 50 or the display processing unit 60). Detailed description regarding the configuration of the data reverse conversion units 52 and 62 will be described later.

Here, processing of image data in the imaging apparatus 1 shown in FIG. 1 will be described. In the shooting operation of the imaging apparatus 1, image data is processed in the following procedure.
(Procedure 1): First, the imaging processing unit 40 performs image data obtained by performing predetermined signal processing on the pixel signal of the subject imaged by the CCD 30 (for example, image data corresponding to the pixel arrangement of the CCD 30). ) Is generated. Then, the generated image data is transferred to the memory 20 via the data conversion unit 41 and the output DMA unit 45 and temporarily stored.
(Procedure 2): Subsequently, the image processing unit 50 reads out the image data temporarily stored in the memory 20 via the input DMA unit 56 and the data reverse conversion unit 52. Then, the image processing unit 50 generates image data (for example, RGB data or YCbCr data in YC422 dot sequential format) obtained by performing recording image processing or display image processing on the read image data. . Thereafter, the generated image data is transferred again to the memory 20 via the data conversion unit 51 and the output DMA unit 55 and temporarily stored.
(Procedure 3): Subsequently, the display processing unit 60 displays image data that has undergone image processing for display by the image processing unit 50, or an OSD (On-Screen Display) display that is separately stored in the memory 20. The image data is read out via the input DMA unit 66 and the data reverse conversion unit 62 and displayed on the display device 70.

  When recording image data captured by the imaging apparatus 1, a recording processing unit (not shown) that performs recording processing of the image data does not show image data that has been subjected to recording image processing by the image processing unit 50. The data is read via the input DMA unit and the data reverse conversion unit, and the read image data for recording is recorded in an image data recording unit (not shown) such as a memory card. For example, when the imaging apparatus 1 has a function of inputting and outputting sound, a voice processing unit (not shown) that performs voice processing converts voice data at the time of shooting into a data conversion unit and an output DMA unit (not shown). Is stored in the memory 20. Further, the audio processing unit reads out the audio data stored in the memory 20 via an input DMA unit and a data reverse conversion unit (not shown), and outputs audio corresponding to the read audio data to a sound output (not shown) such as a speaker. To output.

<First configuration>
Next, the data converter in the imaging device 1 will be described. As described above, the data conversion unit 41 and the data conversion unit 51 are different only in the processing block at the preceding stage and the processing block at the subsequent stage to be connected. More specifically, as shown in FIG. 1, the data conversion unit 41 is arranged between the upstream imaging processing unit 40 and the downstream output DMA unit 45, and the data conversion unit 51 includes the upstream captured image. Arranged between the processing unit 50 and the output DMA unit 55 in the subsequent stage. Accordingly, the data conversion unit 41 and the data conversion unit 51 differ only in the data format (format) of input image data and output image data. In the following description, the data conversion unit 41 will be described as a representative.

  FIG. 2 is a block diagram showing a schematic configuration of the data conversion unit 41 of the first configuration provided in the imaging apparatus 1 of the present embodiment. As shown in FIG. 2, the data conversion unit 41 includes a pack area 1 data generation unit 411, a pack area 2 data generation unit 412, and a data combination unit 413.

  In the following description, Bayer array image data (hereinafter referred to as “Bayer data”) of 16 rows × 16 columns as shown in FIG. A case will be described in which the unit for packing each pixel data is 4 transfers, and 4 transfers is a unit for one burst transfer, that is, 1 burst is a unit for packing pixel data. In the following description, it is assumed that the number of bits of each pixel data in the Bayer data is 11 bits, and the bus width (memory bus width) of the memory bus is 32 bits. In the Bayer data shown in FIG. 15, each pixel data number represents the position of the pixel in the CCD 30, and “R”, “G”, and “B” before the number represent the color of the pixel in the CCD 30. Represents. More specifically, “R” represents pixel data of the red pixel of the CCD 30, “G” represents pixel data of the green pixel of the CCD 30, and “B” represents pixel data of the blue pixel of the CCD 30.

  The data conversion unit 41 arranges each bit data of the 11-bit pixel data input from the imaging processing unit 40 in each bit on the memory bus by any one of predetermined packing methods described later. 32-bit pack data is output. The data conversion unit 41 has a 22-bit area (hereinafter referred to as “pack area 1”) in which each successive bit of pixel data is arranged as it is in each successive bit on the memory bus, and pixel data is arranged. The 10 bits area (hereinafter referred to as “pack area 2”) in which each bit of the pixel data is divided and arranged into the remaining bits on the memory bus, and the bit of the pixel data is assigned to each bit on the memory bus. Arrange each bit.

  Note that the pixel data arranged in the pack area 1 and the pixel data arranged in the pack area 2 in the data conversion unit 41 are determined based on a data enable signal input from the imaging processing unit 40. This data enable signal is a signal that includes information indicating the color data of the pixel of the CCD 30 that is input to the data converter 41.

  The pack area 1 data generation unit 411 is based on the data enable signal input from the imaging processing unit 40, and the pack area 1 data bus (this data bus corresponds to each bit of the memory bus. In this example, pixel data for two pixels is sequentially arranged in each bit of “memory bus” (for example, from a bit corresponding to a lower bit of the memory bus). Then, the pixel data of the packed area 1 (hereinafter referred to as “packed area 1 data”) having a bit number (22 bits in FIG. 2) less than the memory bus width in which the pixel data is sequentially arranged is output to the data combining unit 413. The pack area 1 data generation unit 411 arranges adjacent pixel data of the same color in the same bit on the memory bus in the data between two successive transfers when the pixel data is burst transferred to the memory 20. Thus, the pack area 1 data is arranged.

  Based on the data enable signal input from the imaging processing unit 40, the pack area 2 data generation unit 412 is based on the data bus of the pack area 2 (this data bus also corresponds to each bit of the memory bus. In this example, pixel data for three pixels is arranged in each bit of “memory bus”. Then, the pixel data (hereinafter referred to as “pack area 2 data”) of the pack area 2 having the number of bits (10 bits in FIG. 2) less than the memory bus width in which the pixel data is arranged is output to the data combination unit 413. The pack area 2 data generation unit 412 arranges the pack area 2 data according to a predetermined packing method described later.

  The data combination unit 413 generates pack data obtained by combining the pack region 1 data input from the pack region 1 data generation unit 411 and the pack region 2 data input from the pack region 2 data generation unit 412. Then, the data combination unit 413 sequentially outputs the generated pack data to the output DMA unit 45. The data combining unit 413 includes a memory unit that temporarily holds, for example, 22-bit packed region 1 data. Based on the data enable signal input from the imaging processing unit 40, the 22-bit packed region 1 data is provided. Then, the 32-bit pack data is generated by matching the timing with the 10-bit pack area 2 data.

  Next, the data reverse conversion unit in the imaging apparatus 1 will be described. Note that, as described above, the data inverse transform unit 52 and the data inverse transform unit 62 are different only in the preceding processing block and the subsequent processing block to be connected. More specifically, as shown in FIG. 1, the data reverse conversion unit 52 is disposed between the input DMA unit 56 in the previous stage and the image processing unit 50 in the subsequent stage, and the data reverse conversion unit 62 is in the previous stage. It is arranged between the input DMA unit 66 and the display processing unit 60 at the subsequent stage. Therefore, the data reverse conversion unit 52 and the data reverse conversion unit 62 differ only in the data format (format) of input image data and output image data. In the following description, the data inverse conversion unit 52 will be described as a representative.

  FIG. 3 is a block diagram illustrating a schematic configuration of the data reverse conversion unit 52 of the first configuration provided in the imaging apparatus 1 of the present embodiment. As shown in FIG. 3, the data reverse conversion unit 52 includes a data division unit 521, a pack area 1 data restoration unit 522, a pack area 2 data restoration unit 523, and a data selection unit 524.

  In the following description, one burst of packed data packed in units of 4 transfers packed by the data conversion unit 41 is input to the data inverse conversion unit 52 via the input DMA unit 56 and input. A case where the pack data is returned to the Bayer data shown in FIG. 15 will be described. Therefore, description will be made assuming that the number of bits of each pixel data in the Bayer data is 11 bits and the memory bus width is 32 bits.

  The data reverse conversion unit 52 converts the data arranged in each bit of the 32-bit packed data input from the input DMA unit 56 into an image by a method opposite to any one of the predetermined packing methods described later. The processing unit 50 outputs 11-bit pixel data returned to the original image data (Bayer data) to be subjected to image processing. In the data reverse conversion unit 52, the continuous bit of the original pixel data is arranged as it is in the continuous bit on the memory bus, and the bit region of the original pixel data is divided by dividing the bit of the original pixel data into the memory bus. Each bit data on the memory bus is distributed to each bit of the original pixel data separately from the 10-bit packed area 2 arranged in the upper bits.

  The original pixel data arranged in the pack area 1 and the original pixel data arranged in the pack area 2 in the data reverse conversion unit 52 are discriminated based on the data enable signal input from the image processing unit 50. The This data enable signal is a signal including information indicating the color pixel data of the CCD 30 as the original pixel data output to the image processing unit 50.

  The data dividing unit 521 divides the pack data into pack area 1 data and pack area 2 data. The data dividing unit 521 then sends the divided pack area 1 data (22-bit pack area 1 data in FIG. 3) to the pack area 1 data restoring unit 522, and pack area 2 data (in FIG. 3, 10-bit pack). Area 2 data) is output to the pack area 2 data restoration unit 523.

  The pack area 1 data restoration unit 522 converts the original pixel data for two pixels sequentially arranged in each bit of the pack area 1 data input from the data division unit 521, for example, on the upper side of the pack area 1 data. The pixel data is divided into bits (11 bits in FIG. 3) and lower bits (11 bits in FIG. 3). Then, the divided pixel data is output to the data selection unit 524.

  Based on the data enable signal input from the image processing unit 50, the pack region 2 data restoration unit 523 is configured to store pixel data for three pixels arranged in each bit of the pack region 2 data input from the data dividing unit 521. Is divided into respective pixel data. Then, the divided pixel data (11 bits in FIG. 3) is output to the data selection unit 524.

  Note that the pack area 2 data restoration unit 523 divides the data into the respective pixel data by a method reverse to a predetermined packing method described later. The pack area 2 data restoration unit 523 includes a memory unit that temporarily holds pack area 2 data for a packing unit (4 transfers), for example, and receives a data enable signal input from the image processing unit 50. Based on this, 11-bit pixel data divided and arranged in the pack area 2 data are aligned and output to the data selection unit 524.

  The data selection unit 524 sequentially selects the pixel data input from the pack area 1 data restoration unit 522 or the pack area 2 data restoration unit 523 based on the data enable signal input from the image processing unit 50, and performs inverse data conversion. The original pixel data divided (unpacked) by the unit 52 is output to the image processing unit 50. The data selection unit 524 includes a memory unit that temporarily holds 11-bit pixel data respectively input from the pack area 1 data restoration unit 522 and the pack area 2 data restoration unit 523, and includes an image processing unit. Based on the data enable signal input from 50, the output timing of the pixel data is matched, and each pixel data is sequentially output to the image processing unit 50.

  Next, a pixel data arrangement (packing) method will be described. In the imaging device 1 of the present embodiment, pack data in which the pixel data input by the data conversion units 41 and 51 is arranged on the memory bus is generated, and the pack data input by the data reverse conversion units 52 and 62 is converted into the original pixel. Return to data. In the following description, each pixel data in the Bayer data shown in FIG. 15 is sequentially input to the data conversion unit 41 and packed into one burst pack data with four transfers as one packing unit. explain. Further, the number of bits of each pixel data in the Bayer data is 11 bits, and the memory bus width is 32 bits.

  Note that the method of returning the packed data to the original pixel data by the data inverse conversion units 52 and 62 is the reverse of the method of packing the pixel data by the data conversion units 41 and 51, and thus detailed description thereof is omitted.

<First data arrangement method>
FIG. 4 is a diagram for explaining the first data arrangement method (packing method) in the data converter 41 having the first configuration. FIG. 4A shows the arrangement of pixel data in the pack data, and FIG. 4B shows the arrangement of each bit of the pixel data in the pack area 2 in the pack data.

  As described above, image data usually has a smaller amount of change in pixel data between adjacent pixels. For this reason, when building packed data, the same bit of the pixel data of adjacent pixels is changed to the same bit of the memory bus in each transfer so that the change amount of each bit of the memory bus is minimized during burst transfer. It is desirable to arrange in.

  Also, the color information represented by each pixel data in the image data varies depending on the image format when the image processing unit 50 performs image processing, such as Bayer, RGB, or YC422 dot sequential. For this reason, when building packed data, pixel data of the same color should be placed in the same bit of the memory bus in each transfer so that the amount of change in each bit of the memory bus is minimized during burst transfer Is desirable.

  In the first packing method, the pack area 1 data generation unit 411 sets adjacent pixel data of the same color in the same bit on the memory bus in the data between two successive transfers of the pack area 1 in the pack data. Pack area 1 data is arranged so as to be arranged.

  In addition, the pack area 2 data generation unit 412 sets each pixel data so that the pixel data is in raster order between four consecutive transfers of the pack area 2 in the pack data that is left without pixel data being arranged. Pack area 2 data is arranged by packing bits.

  More specifically, the pack area 1 data generation unit 411 performs pixel data (“R0”, “G1”, “R2”, “G3”, “R4”, “G5”, “R6”) in the Bayer data. , “G7”), as shown in FIG. 4A, “R0” and “G1” in the first transfer, “R2” and “G3” in the second transfer, and “R4” in the third transfer. "G5" and "R6" and "G7" are sequentially arranged (packed) from the least significant bit of the memory bus in the fourth transfer.

  Further, the pack area 2 data generation unit 412 converts each pixel data (“R8”, “G9”, “R10”) in the Bayer data into “R8” at the first transfer as shown in FIG. Bit [0] to bit [9] of bit "10" of "R0" in the second transfer and bit [0] to bit [8] of "G9" in the second transfer, bit [9] of "G9" in the third transfer ] To bit [10] and bit [0] to bit [7] of “R10”, and bits [8] to [10] of “R10” are respectively assigned to the lower bits of the memory bus in the pack area 2 in the fourth transfer. Pack sequentially.

  If all the bits of the pixel data cannot be arranged in the pack area 2, the next pixel data (for example, the pixel data “G11” in the Bayer data) is not packed. This is because the data conversion unit 41 packs each pixel data in a packing unit, and therefore, if a part of the pixel data “G11” is packed, the data “G11” straddles the packing unit. is there. Therefore, the bits of the data bus in the pack area 2 where the pixel data is not arranged (in FIG. 4, bits [31] to [25] of the data bus for the fourth transfer) are unused bits. Note that it is possible to adopt a configuration in which all the data having the same value (for example, all “1” or all “0” data) is arranged in the unused bits.

  By generating the pack data in this way, in the first packing method, the shift of the bit position of the pixel data between each transfer is limited to the bits of the pack area 2 in the entire memory bus width. The Further, in the first packing method, the pixel data in the pack area 1 is arranged so that adjacent pixel data of the same color is arranged in the same bit on the memory bus. Thereby, the amount of change of each bit of the memory bus at the time of burst transfer can be reduced. Thus, in the first packing method, the transfer efficiency of image data is maintained as in the conventional packing method in burst units, and the transfer of image data is more effective than the conventional packing method in burst units. Such power consumption can be reduced.

<Second data arrangement method>
FIG. 5 is a diagram for explaining the second data arrangement method (packing method) in the data conversion unit 41 of the first configuration. FIG. 5A shows the arrangement of the pixel data in the pack data, and FIG. 5B shows the arrangement of each bit of the pixel data in the pack area 2 in the pack data.

  In the second packing method, the pack area 1 data generation unit 411 uses adjacent pixel data of the same color in the data between two successive transfers of the pack area 1 in the pack data, as in the first packing method. However, the pack area 1 data is arranged so as to be arranged in the same bit on the memory bus. Note that the arrangement method of pack data in the pack area 1 in the second packing method is the same as that in the first packing method, and thus detailed description thereof is omitted.

  In addition, the pack area 2 data generation unit 412 divides the pixel data with the same bit position as a boundary in the pack area 2 in the pack data that remains without pixel data being arranged. Then, the pack area 2 data is arranged so that adjacent pixel data of the same color adjacent to the divided pixel data is arranged in the same bit on the memory bus between two successive transfers. The remaining divided bits are arranged at the fourth transfer.

  More specifically, the pack area 2 data generation unit 412 transfers each pixel data (“R8”, “G9”, “R10”) in the Bayer data for one transfer as shown in FIG. Bit [1] to Bit [10] of “R8” in the eye, Bit [1] to Bit [10] of “R10” in the second transfer, Bit [1] to Bit [10] of “G9” in the third transfer Are sequentially packed from the lower bits of the memory bus in the pack area 2. In the fourth transfer, bit [0] of “R8”, bit [0] of “G9”, and bit [0] of “R10” are sequentially packed from the lower bits of the memory bus in the pack area 2, respectively.

  When all the bits of the pixel data cannot be arranged in the pack area 2, as in the first packing method, the bits of the data bus in the pack area 2 where the pixel data is not arranged (see FIG. 5, bit [31] to bit [25]) of the fourth transfer data bus are unused bits.

  In this way, in the second packing method, the same bit of each pixel data is arranged in the same bit on the memory bus also in the pack area 2 (see the first to third transfers in FIG. 5). . Further, adjacent pixel data of the same color are arranged between two successive transfers (see the first transfer and the second transfer in FIG. 5). The bits of the pixel data that are not arranged in the first transfer to the third transfer are arranged in the fourth transfer. Thereby, the amount of change of each bit of the memory bus during burst transfer, in particular, the amount of change of each bit of the pack area 2 can be reduced more than in the first packing method. As a result, in the second packing method, it is possible to further reduce the power consumption related to the transfer of the image data, compared to the first packing method.

  In the second packing method shown in FIG. 5, the case has been described in which the bit [0] of each pixel data is collectively arranged when the pixel data is arranged in the pack area 2. This is because the memory bus width of the pack area 2 is 10 bits and the number of bits of the pixel data is 11 bits. That is, when pixel data is arranged on the memory bus, the least significant 1 bit of each pixel data that could not be arranged on the memory bus in one transfer (in FIG. 5, bit [0] of each pixel data) Are arranged together in the fourth transfer. Therefore, when there are a plurality of bits of pixel data that could not be arranged on the memory bus in one transfer, the pixel data of the plurality of bits are arranged together at the fourth transfer. Note that the bits of the pixels arranged in the fourth transfer are preferably lower bits. This is because the image data is normally expected to have different values depending on noise components in the imaging device 1, for example, even if the lower bits of the pixel data are adjacent to each other.

<Third data arrangement method>
FIG. 6 is a diagram for explaining the third data arrangement method (packing method) in the data conversion unit 41 of the first configuration. FIG. 6A shows the arrangement of pixel data in the pack data, and FIG. 6B shows the arrangement of each bit of the pixel data in the pack area 2 in the pack data.

  In the third packing method, the pack area 1 data generation unit 411 is adjacent to the data between two consecutive transfers in the pack area 1 in the pack data, as in the first packing method and the second packing method. The pixel data is arranged so that pixel data of the same color is arranged at the same bit on the memory bus. Note that the arrangement method of the pack data in the pack area 1 in the third packing method is the same as the first packing method and the second packing method, and thus detailed description thereof is omitted.

  Also, the pack area 2 data generation unit 412 outputs pixel data with the same bit position as a boundary in the pack area 2 in the pack data remaining without the pixel data being arranged, as in the second packing method. Divide and arrange. Further, in the third packing method, the pack area 2 data generation unit 412 again arranges the pixel data having the same bits as the pixel data arranged in the previous one transfer in the unused bits in the pack area 2, that is, Copy and place. Note that the arrangement method of pack data other than unused bits in the pack area 2 in the third packing method is the same as that in the second packing method, and thus detailed description thereof is omitted.

  Here, the arrangement of pixel data of unused bits by the pack area 2 data generation unit 412 will be described more specifically. The pack area 2 data generation unit 412 uses the unused bits of the fourth transfer (in FIG. 6, bits [31] to [25] of the data bus of the fourth transfer) to the bits [ 31] to bit [25], the pixel data (bit [4] to bit [10] of pixel data “G9” in FIG. 6) is copied and rearranged.

  In this way, in the third packing method, the same value as the pixel data arranged in the previous one transfer is arranged in the unused bits in the pack area 2 (see the fourth transfer in FIG. 6). . As a result, the amount of change of each bit of the memory bus during burst transfer, in particular, the amount of change of unused bits in the pack area 2 can be further reduced as compared with the second packing method. Thus, in the third packing method, it is possible to further reduce the power consumption related to the transfer of image data, compared to the second packing method.

  As described above, in the data conversion units 41 and 51 of the first configuration, the pixel data can be arranged so that the change amount of each bit of the memory bus at the time of burst transfer is reduced. Thereby, in the data conversion units 41 and 51 of the first configuration, the image data transfer efficiency is maintained as in the conventional burst unit packing method, and the conventional burst unit packing method, It is possible to reduce power consumption related to transfer of image data. As a result, power consumption related to transfer of pixel data between each component (processing block) in the imaging device 1 and the memory 20, that is, transfer of pixel data between each component in the imaging device 1. Can be reduced.

<Second configuration>
Next, the data conversion unit having the second configuration will be described. FIG. 7 is a block diagram illustrating a schematic configuration of the data conversion unit 43 of the second configuration provided in the imaging apparatus 1 of the present embodiment. The data conversion unit 43 illustrated in FIG. 7 is provided in the imaging device 1 instead of the data conversion unit 41 having the first configuration. Therefore, the data conversion unit 43 can be provided in the imaging device 1 instead of the data conversion unit 51. As shown in FIG. 7, the data conversion unit 43 includes a bit division unit 430, a pack region 1 data generation unit 431, a pack region 2 data generation unit 432, and a data combination unit 433.

  In the following description, as in the first configuration, Bayer data of 16 rows × 16 columns as shown in FIG. 15 is input to the data conversion unit 43, and four transfers are performed in each Bayer data. A case where the pixel data packing unit is used and the memory bus width is 32 bits will be described. In FIG. 7, unlike the second configuration, description will be made assuming that the number of bits of each pixel data in the Bayer data is 9 bits.

  The data converter 43 outputs 32-bit pack data in which each bit of 9-bit pixel data input from the imaging processor 40 is arranged in each bit on the memory bus by a predetermined packing method described later. To do. Similar to the data converter 41 of the first configuration, the data converter 43 divides the packing unit into the pack area 1 and the pack area 2 and arranges each bit of the pixel data in each bit on the memory bus. However, the method of arranging each bit of pixel data in each bit on the memory bus is different from the data conversion unit 41 of the first configuration. A detailed description of the arrangement method of the pixel data on the memory bus by the data conversion unit 43 will be described later.

  The bit division unit 430, for example, converts each input pixel data into upper 8 bit pixel data (hereinafter referred to as “upper pixel data”) and lower 1 bit pixel data (hereinafter referred to as “lower pixel data”). And split. Then, the divided upper pixel data is output to the pack area 1 data generation unit 431. In addition, the bit division unit 430 outputs the divided lower pixel data to the pack region 2 data generation unit 432.

  The pack area 1 data generation unit 431 is based on the data enable signal input from the imaging processing unit 40, and the pack area 1 data bus (this data bus corresponds to each bit of the memory bus. In this example, upper pixel data for four pixels is sequentially arranged in each bit of “memory bus”, for example, starting from a bit at a position corresponding to the lower bit of the memory bus. Then, the pack area 1 data having the number of bits (32 bits in FIG. 7) of the memory bus width in which the upper pixel data is sequentially arranged is output to the data combination unit 433. When the pack area 1 data generation unit 431 outputs the pack area 1 data for the fourth transfer to the data combination unit 433, the upper pixel data for two pixels corresponds to, for example, the lower bits of the memory bus. The 16 bit packed region 1 data arranged sequentially from the position bit is output to the data combining unit 433. The pack area 1 data generation unit 431 arranges adjacent pixel data of the same color in the same bit on the memory bus in the data between two successive transfers when the upper pixel data is burst transferred to the memory 20. Thus, the pack area 1 data is arranged.

  The pack area 2 data generation unit 432 is based on the data enable signal input from the imaging processing unit 40, and the pack area 2 data bus (this data bus also corresponds to each bit of the memory bus. In FIG. 4, the lower pixel data for 14 pixels is arranged in each bit of “memory bus”. Then, the pack area 2 data having the number of bits less than the memory bus width in which the lower pixel data is arranged (16 bits in FIG. 7) is output to the data combining unit 433. The pack area 2 data generation unit 432 includes, for example, a memory unit that temporarily holds 16 pieces of 1-bit pixel data, and pixel data based on a data enable signal input from the imaging processing unit 40. The pack area 2 data is output to the data combining unit 433 at the timing of the fourth transfer.

  The data combining unit 433 generates pack data obtained by combining the pack region 1 data input from the pack region 1 data generation unit 431 and the pack region 2 data input from the pack region 2 data generation unit 432. Then, the data combining unit 433 sequentially outputs the generated pack data to the output DMA unit 45. The data combining unit 433 sequentially outputs the pack area 1 data input from the pack area 1 data generation unit 431 to the output DMA unit 45 as pack data from the first transfer to the third transfer. Also, the data combining unit 433 receives the 16-bit pack region 1 data input from the pack region 1 data generation unit 431 and the 16-bit pack region 2 data input from the pack region 2 data generation unit 432 in the fourth transfer. Is output to the output DMA unit 45.

  The data combining unit 433 includes, for example, a memory unit that temporarily stores 16-bit packed area 1 data. Based on the data enable signal input from the imaging processing unit 40, the fourth transfer of 16-bit data is performed. The pack area 1 data and the 16-bit fourth transfer pack area 2 data are matched in timing to generate 32-bit pack data.

  Next, the data inverse conversion unit having the second configuration will be described. FIG. 8 is a block diagram showing a schematic configuration of the data inverse conversion unit 54 of the second configuration provided in the imaging apparatus 1 of the present embodiment. The data reverse conversion unit 54 illustrated in FIG. 8 is provided in the imaging device 1 instead of the data reverse conversion unit 52 having the first configuration. Therefore, the data reverse conversion unit 54 can be provided in the imaging device 1 instead of the data reverse conversion unit 62. As shown in FIG. 8, the data reverse conversion unit 54 includes a data division unit 541, a pack area 1 data restoration unit 542, a pack area 2 data restoration unit 543, a bit combination unit 540, and a data selection unit 544. It is equipped with.

  In the following description, as in the first configuration, one burst of packed data packed in the data inverse conversion unit 54 by the data conversion unit 43 and having 4 transfers as a packing unit is input DMA unit 56. A case where the input pack data is returned to the Bayer data shown in FIG. 15 will be described. Therefore, description will be made assuming that the number of bits of each pixel data in the Bayer data is 9 bits and the memory bus width is 32 bits.

  The data reverse conversion unit 54 converts the data arranged in each bit of the 32-bit packed data input from the input DMA unit 56 by the image processing unit 50 using a method reverse to a predetermined packing method described later. The 9-bit pixel data returned to the original image data (Bayer data) is output. Similar to the data inverse conversion unit 52 of the first configuration, the data inverse conversion unit 54 divides the packing unit into the pack area 1 and the pack area 2, and converts the data of each bit on the memory bus to the original pixel. Distribute to each bit of data.

  The data dividing unit 541 divides the pack data into pack area 1 data and pack area 2 data. Then, the data division unit 541 outputs the divided pack area 1 data (32-bit pack area 1 data or 16-bit pack area 1 data in FIG. 8) to the pack area 1 data restoration unit 542. Further, the data dividing unit 541 outputs the pack area 2 data (16-bit pack area 2 data in FIG. 8) to the pack area 2 data restoring unit 543.

  The pack area 1 data restoration unit 542 converts the original upper pixel data for 4 pixels sequentially arranged in each bit of the pack area 1 data input from the data division unit 541 into, for example, four pack area 1 data The data is divided into upper pixel data (8 bits in FIG. 8). Then, each divided upper pixel data is output to the bit combination unit 540.

  Based on the data enable signal input from the image processing unit 50, the pack region 2 data restoration unit 543 is arranged for each of the lower 14 pixels arranged in each bit of the pack region 2 data input from the data dividing unit 541. Data is divided into respective pixel data. Then, each divided lower pixel data (1 bit in FIG. 8) is sequentially output to the bit combination unit 540. Note that the pack area 2 data restoration unit 543 divides the data into lower pixel data by a method reverse to a predetermined packing method described later.

  The bit combination unit 540 combines the upper pixel data input from the pack area 1 data restoration unit 542 and the lower pixel data input from the pack area 2 data restoration unit 543, and the combined pixel data is a data selection unit. Output to 544. The bit combination unit 540 includes, for example, a memory unit that temporarily stores upper pixel data for a packing unit (4 transfers), and each time lower pixel data is input from the pack region 2 data restoration unit 543. In addition, the upper pixel data corresponding to the inputted lower pixel data is combined, and 9-bit pixel data arranged by being divided into the pack area 1 data and the pack area 2 data is aligned, and the data selection unit 544 Output.

  The data selection unit 544 sequentially selects the pixel data input from the bit combination unit 540 based on the data enable signal input from the image processing unit 50, and the original data divided (unpacked) by the data inverse conversion unit 54 The pixel data is output to the image processing unit 50. The data selection unit 544 includes a memory unit that temporarily stores 9-bit pixel data input from the bit combination unit 540, respectively, and is based on a data enable signal input from the image processing unit 50. The output timing of the pixel data is matched, and each pixel data is sequentially output to the image processing unit 50.

  Next, a pixel data arrangement (packing) method in the second configuration will be described. In the imaging apparatus 1 of the present embodiment, for example, pack data in which the pixel data input by the data conversion unit 43 is arranged on a memory bus is generated, and the pack data input by the data reverse conversion unit 54 is converted into original pixel data. return. In the following description, each pixel data in the Bayer data shown in FIG. 15 is sequentially input to the data conversion unit 43 and packed into one burst pack data with four transfers as one packing unit. explain. In addition, the number of bits of each pixel data in the Bayer data is 9 bits, and the memory bus width is 32 bits.

  Note that the method for returning the packed data to the original pixel data by the data reverse conversion unit 54 is the reverse of the method for packing the pixel data by the data conversion unit 43, and thus detailed description thereof is omitted.

<Fourth data arrangement method>
FIG. 9 is a diagram for explaining the fourth data arrangement method (packing method) in the data conversion unit 43 of the second configuration. FIG. 9A shows an example in which 9-bit pixel data is arranged on the memory bus by the above-described second packing method, and FIG. 9B shows the contents of the pack data by the fourth packing method. FIG. 9C shows the arrangement of each bit of the pixel data in the pack area 2 in the pack data according to the fourth packing method.

  In the fourth packing method, in the data between two consecutive transfers in the pack area 1 in the pack data, the pack area 1 data is arranged such that adjacent pixel data of the same color are arranged in the same bit on the memory bus. Place. Also, within the packed area 2 in the packed data, the pixel data is divided with the same bit position as a boundary, and adjacent pixel data of the same color adjacent to the divided pixel data are the same on the memory bus between two successive transfers. Pack area 2 data is arranged so as to be arranged in bits.

  Therefore, in the second packing method, as shown in FIG. 9A, each pixel data (“R0”, “G1”, “R2”, “G3”, “R4”, “G5”, “R6”, “G7”, “R8”, “R10”, “R12”) are arranged in the pack area 1, and each pixel data (“G9”, “G11”, “ G13 ") is arranged in the pack area 2.

  Even in the second packing method shown in FIG. 9A, the amount of change in each bit of the memory bus at the time of burst transfer is reduced, so that the power consumption related to the transfer of image data can be reduced. It is expected that the power consumption will increase with bits [27] to [31] of the memory bus in the third transfer and bits [19] to [26] of the memory bus in the fourth transfer.

  In the image data, when the change amount of the pixel data between adjacent pixels is small, it is expected that the upper bits of the pixel data do not change and only the lower bits change. For this reason, the pack data is constructed by dividing the pixel data into upper bits and lower bits, placing upper bits with a small amount of change in the pack area 1, and placing lower bits with a large amount of change in the pack area 2. However, the amount of change of each bit of the memory bus during burst transfer may be further reduced.

  In the fourth packing method, the bit dividing unit 431 divides pixel data into upper pixel data and lower pixel data. Then, the pack area 1 data generation unit 431 uses the divided upper pixel data in the data between two successive transfers of the pack area 1 in the pack data, and the adjacent upper pixel data of the same color is the same on the memory bus. Pack area 1 data is arranged so as to be arranged in bits. The pack area 2 data generation unit 432 arranges the divided lower pixel data at the fourth transfer.

  More specifically, the bit dividing unit 431 divides each pixel data in the Bayer data into upper pixel data (bit [8] to bit [1]) and lower pixel data (bit [0]). . Then, as shown in FIG. 9B, the pack area 1 data generation unit 431 performs “R0”, “G1”, “R8”, and “G9” for the first transfer, and “R2” for the second transfer. , “G3”, “R10”, and “G11”, “R4”, “G5”, “R12”, and “G13” in the third transfer, and “R6” and “G7” upper pixel data in the fourth transfer Are sequentially arranged (packed) from the least significant bit of the memory bus.

  The pack area 2 data generation unit 432 also performs “R0”, “G1”, “R2”, “G3”, “R4”, “G5”, “R6”, “G7”, “R8”, “G9”. , “R10”, “G11”, “R12”, and “G13” are sequentially packed from the lower bits of the memory bus in the pack area 2, respectively. Note that bit [30] and bit [31] of the memory bus in the fourth transfer are unused bits. This is the same reason as in the first configuration.

  By generating the pack data in this way, in the fourth packing method, as in the first configuration, the shift of the bit position of the pixel data between the transfers is out of the entire memory bus width. It is limited to the bits in the pack area 2. Further, in the fourth packing method, the pack area 2 is limited to only the fourth transfer, and therefore the ratio of the pack area 2 in the memory bus can be reduced. Thereby, the amount of change of each bit of the memory bus at the time of burst transfer can be further reduced. Thus, in the fourth packing method, it is possible to further reduce the power consumption related to the transfer of the image data, compared to the packing method (first to third packing methods) of the first configuration.

  In the fourth packing method, similarly to the third packing method, the pixels arranged in the unused bits shown in FIG. 9 in bits [30] to [31] of the third transfer data bus. By copying the data (bit [7] to bit [8] of the pixel data “G13” in FIG. 9) and arranging the data again, the power consumption related to the transfer of the image data can be further reduced. .

  As described above, in the data conversion unit 43 of the second configuration, the pixel data can be arranged so that the change amount of each bit of the memory bus at the time of burst transfer is reduced. As a result, the data conversion unit 43 of the second configuration also maintains the transfer efficiency of the image data in the same manner as the conventional packing method in the burst unit, and also the image data than the conventional packing method in the burst unit. It is possible to reduce the power consumption related to the transfer. Accordingly, similarly to the data conversion units 41 and 51 of the first configuration, the pixel data is transferred between each component (processing block) in the imaging device 1 and the memory 20, that is, in the imaging device 1. Thus, it is possible to reduce the power consumption related to the transfer of pixel data between the components.

  In the embodiment described above, the case where the packing method of the present embodiment is applied to Bayer data has been described, but the packing method of the present embodiment is not limited to application to Bayer data, It can be applied to various image data such as RGB data, YC422 dot sequential or YC444 dot sequential format YCbCr data, or OSD data. Further, the packing method of the present embodiment is not limited to application to image data, and can also be applied to other types of data such as audio data.

<Application example>
Next, an example in which the packing method of the present embodiment is applied to other types of image data or data other than image data will be described. First, an example in which the packing method of the present embodiment is applied to RGB data will be described. FIG. 10 is a diagram illustrating an example when the data arrangement method (packing method) by the data conversion unit 41 included in the imaging apparatus 1 of the present embodiment is applied to other image data (RGB data). In FIG. 10, an example in which the conventional packing method in units of bursts is applied to 8-bit RGB data (see FIG. 10A) and an example in which the third packing method is applied (see FIG. 10B). ), Respectively.

  As shown in FIG. 10A, when RGB data (8 bits × 3 colors = 24 bits) is simply arranged by a conventional packing method in units of bursts, all bits of a 32-bit memory bus have different colors. It becomes pixel data. On the other hand, in the third packing method shown in FIG. 10B, each bit of the memory bus in the packed area 1 becomes the same bit of the pixel data of the same color for the four transfers. Further, the value of each bit of the memory bus for the fourth transfer in the pack area 2 is the same as the value of each bit of the memory bus for the third transfer. Thus, in the pack area 1, the amount of change in each bit of the memory bus during burst transfer can be reduced, and in the pack area 2, the amount of change in each bit of the memory bus between the third transfer and the fourth transfer can be reduced. Can be eliminated. As a result, power consumption related to transfer of image data can be reduced.

  FIG. 11 is a diagram illustrating an example when the data arrangement method (packing method) by the data conversion unit 41 included in the imaging apparatus 1 of the present embodiment is applied to still another image data (YCbCr data). FIG. 11 shows an example in which a conventional packing method in units of bursts is applied to 10-bit Y, 8-bit Cb and Cr data as shown in FIG. 11A (see FIG. 11B). FIG. 11 shows an example of applying the first packing method (see FIG. 11C).

  As shown in FIG. 11B, if YCbCr data is simply arranged by the conventional packing method in units of bursts, all the bits of the memory bus become pixel data representing different colors. On the other hand, in the first packing method shown in FIG. 11C, each bit of the memory bus in the packed area 1 is the same bit of pixel data representing the same color. Thereby, in the pack area 1, the amount of change of each bit of the memory bus at the time of burst transfer can be reduced, and power consumption related to transfer of image data can be reduced. As can be seen from the example shown in FIG. 11, the packing method of this embodiment can be applied in the same manner even when the number of bits of each data is different.

  FIG. 12 is a diagram illustrating an example when the data arrangement method (packing method) by the data conversion unit 41 provided in the imaging apparatus 1 of the present embodiment is applied to still another image data (OSD data). In FIG. 12, an example in which the conventional packing method in units of bursts is applied to 6-bit OSD data (see FIG. 12A) and an example in which the second packing method is applied (see FIG. 10B). ), Respectively.

  As shown in FIG. 12A, if the OSD data is simply arranged by the conventional packing method in units of bursts, all bits of the memory bus become different bits of different display data. On the other hand, in the second packing method shown in FIG. 12B, each bit of the memory bus in the packed area 1 becomes the same bit of display data. Thereby, in the pack area 1, the amount of change of each bit of the memory bus at the time of burst transfer can be reduced, and power consumption related to transfer of image data can be reduced. Also, as can be seen from FIG. 12B, in the packing method of this embodiment, since most of the 32-bit memory bus width is the pack area 1, it is possible to obtain more power consumption reduction effect. .

  FIG. 13 is a diagram illustrating an example when the data arrangement method (packing method) by the data conversion unit 41 included in the imaging apparatus 1 of the present embodiment is applied to other data (voice data). FIG. 13 shows a case where 24-bit R (right) data and L (left) data are applied to a 64-bit memory bus. FIG. 13A shows an example in which the conventional packing method in units of bursts is applied, and FIG. 13B shows an example in which the second packing method is applied.

  As shown in FIG. 13A, if audio data is simply arranged by the conventional packing method in units of bursts, all the bits of the 64-bit memory bus become audio data of different bits. On the other hand, in the second packing method shown in FIG. 13B, each bit of the memory bus in the pack area 1 becomes the same bit of the audio data on the same side. Thereby, in the pack area 1, the amount of change of each bit of the memory bus at the time of burst transfer can be reduced, and the power consumption related to transfer can be reduced even for other types of data other than image data.

  As described above, according to the embodiment for carrying out the present invention, the memory bus width is divided into the pack area 1 and the pack area 2. Then, each bit of the data to be transferred is arranged almost as it is in each bit on the memory bus in the pack area 1. In addition, each bit of the data to be transferred is divided and arranged in each bit on the memory bus in the pack area 2. Thereby, the amount of change of each bit of the memory bus at the time of burst transfer can be reduced. In addition, the data transfer efficiency can maintain the same data transfer efficiency as before. Thereby, it is possible to sufficiently obtain the effect of reducing the power consumption related to the data transfer between each component (processing block) in the image apparatus and the memory.

  In this embodiment, the case where four transfers are used as a unit of burst transfer and one burst is used as a data packing unit has been described. However, the number of transfers in one burst transfer and the number of data packings A unit is not limited to the form for implementing this invention. For example, 64 transfers may be used as a unit for one burst transfer, and 16 transfers within one burst may be used as a data packing unit, that is, one packing may include four packing units. Further, for example, 16 transfers may be used as a data packing unit and 8 transfers may be used as one burst transfer unit, that is, the packing unit may extend over two burst transfers.

  In the present embodiment, the case where the packing method in the case of transferring data between components in the imaging apparatus is applied to some data has been described. However, the packing method of the present embodiment is applied. The range in which data can be transmitted is not limited to the mode for carrying out the present invention, and can be applied to any system as long as it is a system that exchanges data between each component.

  In the present embodiment, for example, the data conversion unit is arranged as a processing block preceding the output DMA unit and the data inverse conversion unit is arranged as a processing block subsequent to the input DMA unit. The arrangement of the data conversion unit and the data reverse conversion unit is not limited to the form for carrying out the present invention. For example, the data conversion unit can be a component in the output DMA unit, and the data reverse conversion unit can be a component in the input DMA unit. Further, for example, the data conversion unit can be a component in the preceding image processing unit, and the data reverse conversion unit can be a component in the subsequent image processing unit.

  The embodiment of the present invention has been described above with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes various modifications within the scope of the present invention. It is.

1 ... Imaging device (data processing device)
10 ... CPU (data processing device)
20 ... Memory (data processing device)
30 ... CCD
40 ... Imaging processor 50 ... Image processor 60 ... Display processor 70 ... Display devices 41, 51 ... Data converter (data processing device)
411... Pack area 1 data generation unit (first data generation unit)
412... Pack area 2 data generation unit (second data generation unit)
413... Data combination unit 52, 62... Data reverse conversion unit (data processing device)
521: Data division unit 522: Pack area 1 data restoration unit (first data restoration unit)
523... Pack area 2 data restoration unit (second data restoration unit)
524 ... Data selection unit 45, 55 ... Output DMA unit 56, 66 ... Input DMA unit 43 ... Data conversion unit (data processing device)
430: Bit division unit (pixel data division unit)
431... Pack area 1 data generation unit (first data generation unit)
432 ... Pack region 2 data generation unit (second data generation unit)
433... Data combining unit 54... Data reverse conversion unit (data processing device)
541... Data division unit 542... Pack area 1 data restoration unit (first data restoration unit)
543... Pack area 2 data restoration unit (second data restoration unit)
540 ... Bit combination part (pixel data combination part)
544 ... Data selection section

Claims (6)

A data processing device for generating pack data for burst transfer of a plurality of input data on a data bus,
By setting a first bus area and a second bus area on the data bus, the pack data is divided into a first pack area and a second pack area, and the first data is taken out, A first data generation unit that packs the first pack area without a gap so that the first data is transferred from the same position in the first bus area during burst transfer of the first data;
A second data generation unit that packs the second data, which is the input data left by the first data generation unit, into the second pack area;
A data combining unit that combines the first data packed in the first pack area and the second data packed in the second pack area to generate the pack data;
With
The plurality of input data is image data having a plurality of color information, and in the data between two successive transfers of the first pack area in the pack data, adjacent pixel data of the same color are on the memory bus. Of the same bit,
A data processing apparatus. The second data generator is
Packing the second data so as to be continuous in the second pack area;
The data processing apparatus according to claim 1. A data processing apparatus for restoring packed original data from packed data burst-transferred on a data bus having a predetermined number of bits,
A first bus area and a second bus area are set on the data bus, the data of the first pack area of the pack data is extracted from the first bus area and output, and the second bus is output. A data dividing unit for extracting and outputting the data of the second pack area of the pack data from the area;
A first data restoration unit that restores the first data based on the data extracted from the first pack area;
A second data restoration unit for restoring second data based on the data extracted from the second pack area;
A data selection unit for outputting original data based on the restored first data and the second data;
With
The first data is:
Data packed in the first packed area without any gaps so as to be transferred from the same position in the first bus area during the burst transfer;
The original data is image data having a plurality of color information. In the data between two successive transfers in the first pack area in the pack data, adjacent pixel data of the same color are the same on the memory bus. Arranged in a bit,
A data processing apparatus. The second data is:
The data is packed continuously in the second packed area.
The data processing apparatus according to claim 3. A data processing method for generating pack data for burst transfer of a plurality of input data on a data bus,
Dividing the pack data into a first pack area and a second pack area by setting a first bus area and a second bus area on the data bus;
Extracting the first data and packing the first pack area without gaps so that the first data is transferred from the same position in the first bus area during burst transfer of the first data;
Packing the second data, which is the input data remaining from the first data, into the second pack area;
Combining the first data packed in the first pack area and the second data packed in the second pack area to generate the pack data;
Including
The plurality of input data is image data having a plurality of color information, and in the data between two successive transfers of the first pack area in the pack data, adjacent pixel data of the same color are on the memory bus. Are arranged in the same bit of
A data processing method. A data processing method for restoring original packed data from packed data burst-transferred on a data bus having a predetermined number of bits,
Setting a first bus area and a second bus area on the data bus;
Extracting and outputting data of the first pack area of the pack data from the first bus area; and
Extracting and outputting data of the second pack area of the pack data from the second bus area; and
Restoring the first data based on the data retrieved from the first pack area;
Restoring the second data based on the data retrieved from the second pack area;
Outputting original data based on the restored first data and the second data;
Including
The first data is:
Data packed in the first packed area without any gaps so as to be transferred from the same position in the first bus area during the burst transfer;
The original data is image data having a plurality of color information. In the data between two successive transfers in the first pack area in the pack data, adjacent pixel data of the same color are the same on the memory bus. Is the data placed in the bits,
A data processing method.

Images