JP2005109606A - Signal processing method, signal processing apparatus, recording apparatus, and reproducing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a signal processing method whereby a compression rate of an input image can be set in response to an object. <P>SOLUTION: An object detection block 90 detects a face image included in received image data in units of frames, and a quantization section 74 of a data processing encode/decode section 31 decreases quantization coefficients of macroblocks including the face image and increases quantization coefficients of the other macroblocks than the macroblocks including the face image when the object detection block 90 detects the face image. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention includes, for example, a signal processing method and a signal processing apparatus suitable for recording photographed image information on a recording medium and reproducing image information recorded on the recording medium, and such a signal processing apparatus. The present invention relates to a recording apparatus and a reproducing apparatus.

  In recent years, a video camera has been proposed that can record a photographed image taken by a user on a recording medium such as an optical disk or reproduce and display an image recorded on the recording medium.

  In video cameras such as those described above, compression processing such as the MPEG (Moving Picture Experts Group) method is performed in order to compress the amount of information when a captured image taken with a camera block is recorded on a disk or the like. . In the case of the MPEG system, motion detection is performed on the image information from the camera block by the motion detection block. Then, DCT (Discrete Cosine Transform) is performed on the image information obtained based on the detection result of the motion vector and the captured image. Further, in order to reduce the amount of information, requantization is performed in the quantization block (high-frequency component is set to 0), and the block order is rearranged so that the block is zigzag from the block at the upper left of the screen of one frame Then, compression encoding is performed by performing run length coding and further compressing the information amount.

  By the way, in the quantization block as described above, quantization is performed based on the quantization coefficient. The smaller the quantization coefficient, the smaller the quantization noise and the higher the image quality. However, if the quantization coefficient is reduced, the compression rate also decreases accordingly, and the amount of information after compression encoding increases. On the contrary, since the compression rate increases as the quantization coefficient is increased, the amount of information after compression coding can be reduced. However, in this case, the quantization noise increases, and the image quality deteriorates.

  The quantization coefficient is usually calculated and determined from motion vector information of the motion detection block. For this reason, for example, when the movement of the photographed image is large, the quantization coefficient is reduced as an important image to improve the image quality. On the contrary, when the motion of the photographed image is small, it is determined that the image is not an important image, and the quantization coefficient is increased to reduce the amount of information. However, since the degree of importance and the amount of motion of the captured image do not necessarily coincide with each other, it is not necessarily appropriate to determine the quantization coefficient using motion vector information.

Therefore, as a method for determining the quantization coefficient, weighted compression coding has been proposed in which a captured image is divided into an important region and other regions, and bit allocation is adaptively performed according to the importance of the image. ing.
As an index to divide an image into important areas and other areas, for example, because human eyes are sensitive to everyday skin color, is the image important by setting a threshold for skin color? A technique for determining whether or not is known. According to such a method, it becomes possible to detect areas such as a face and a hand included in a moving image, and to assign bits preferentially to macroblocks corresponding to these areas. Thus, it is possible to improve the image quality viewed from the visual sense.

  In Patent Document 1 below, a speaker is specified from the characteristics of an audio signal, a speaker is specified from movements of the face and mouth, and after a predetermined period has elapsed after recording audio information and video information, or stored. A technology for changing the compression rate of audio information and video information based on the information of the section specified from the audio signal when the used capacity of the medium exceeds a predetermined value or the free capacity becomes less than the predetermined value Has been proposed.

JP-A-10-214270

However, when human skin color as described above is detected and the image is divided into important areas and other areas, the separation method when the background color is similar to the skin color, or the There was a need to improve the robustness against change and race.
Also, for example, when the amount of information assigned to an image is changed depending on the importance of the object included in the image, or when only a specific object is to be recorded, it cannot be realized by image separation based on skin color detection as described above. there were.

  In speaker recognition using the audio signal described in Patent Document 1, the compression rate of image information can be changed, for example, in units of frames. It was difficult to separate these into important areas and other areas.

  Therefore, the present invention has been made in view of the above points, and the signal processing method of the present invention detects a specific object included in input image information, and when a specific object is detected. At least a specific object included in the input image information is compressed by changing the compression rate.

  The signal processing device of the present invention includes an object detection unit that detects a specific object included in the input image information, and a specific object included in at least the input image information when the specific object is detected by the object detection unit. Is provided with an image compression means for compressing by changing the compression rate.

  According to the present invention as described above, a specific object included in the input image information is detected, and when the specific object is detected, at least the specific object included in the input image information is changed in compression rate. I am doing so. Thereby, the compression rate of the input image information can be changed according to the object.

  Further, the recording apparatus of the present invention includes an object detection unit that detects a specific object included in input image information, a storage unit that stores information on a required object, and a specific object detected by the object detection unit. The matching means for checking whether the object matches the required object stored in the storage means, and the specific object detected by the object detection means based on the verification result of the matching means is stored in the storage means. And a recording control means for performing recording control for recording the input image information including the specific object on the recording medium when it is determined that the object matches.

  According to the recording apparatus of the present invention as described above, whether or not the object detected from the input image information by the object detection means matches the required object stored in the storage means is verified by the verification means. Based on the result, the recording control means performs recording control for recording the input image information on the recording medium. This makes it possible to perform recording control for recording input image information on a recording medium in accordance with the object detected by the object detection means.

  Further, the reproducing apparatus of the present invention includes a reading unit that reads image information from a recording medium, a decoding unit that performs predetermined decoding processing on the image information read by the reading unit and outputs the image information as image information, Object detection means for detecting a specific object from the image information output from the decoding means, and reading control means for controlling the reading operation of the image information from the recording medium by the reading means based on the detection result of the object detection means. I decided to prepare.

  According to the playback apparatus of the present invention as described above, the object detection unit detects an object from the image information from the recording medium, and based on the detection result, the reading unit reads the image information from the recording medium. Control is performed. Thereby, it is possible to control the reading operation of the image information from the recording medium in accordance with the object detected by the object detecting means.

As described above, the signal processing method and the signal processing apparatus of the present invention detect a specific object included in the input image information, and at least a specific object included in the input image information when the specific object is detected. The compression rate is changed. For example, image information including an important object is compressed at a compression rate smaller than that of image information in other areas and output.
This makes it possible to reduce the total amount of image information while ensuring high image quality for important object images.

  Further, the recording apparatus of the present invention collates whether or not the object detected from the input image information in the object detection unit matches the required object stored in the storage unit, and based on the collation result, The recording control means performs recording control for recording the input image information on the recording medium. This makes it possible to perform recording control for recording input image information on a recording medium in accordance with the object detected by the object detection means. For example, it is possible to record only input image information including an important object on the recording medium, and it is possible to effectively use the recording capacity of the recording medium.

  In the reproducing apparatus of the present invention, the object detection means detects an object from the image information from the recording medium, and controls the reading operation of the image information from the recording medium by the reading means based on the detection result. ing. As a result, it becomes possible to control the reading operation of the image information from the recording medium according to the object detected by the object detecting means. For example, the image information including the required object is searched from the recording medium. Can be read out.

Hereinafter, embodiments of the present invention will be described.
In this embodiment, a video camera in which a camera device unit and a recording / reproducing device unit capable of recording / reproducing images (still images or moving images) and audio are given as an example.
The description will be given in the following order.
1. 1. Configuration of video camera 2. Configuration example of encoding block 3. Structure of object detection block 4. Other configuration examples of the encoding block 5. Decoding block configuration example Configuration example of speaker recognition unit

1. Video camera configuration

FIG. 1 is a block diagram illustrating a configuration example of a video camera according to the present embodiment.
In the lens block 1 shown in this figure, for example, an optical system 11 configured with an imaging lens, a diaphragm, and the like is actually provided. The lens block 1 includes a focus motor for causing the optical system 11 to perform an autofocus operation, a zoom motor for moving the zoom lens based on the operation of the operation unit 7, and the like as the motor unit 12. .

The camera block 2 is provided with a circuit unit for mainly converting image light photographed by the lens block 1 into a digital image signal.
The CCD (Charge Coupled Device) 21 of the camera block 2 is given an optical image of the subject that has passed through the optical system 11. The CCD 21 performs photoelectric conversion on the optical image to generate an imaging signal and supplies it to a sample hold / AGC (Automatic Gain Control) circuit 22. The sample hold / AGC circuit 22 performs gain adjustment on the image pickup signal output from the CCD 21 and performs waveform shaping by performing sample hold processing. The output of the sample hold / AGC circuit 2 is supplied to the video A / D converter 23 to be converted into digital image data.

  Signal processing timing in the CCD 21, sample hold / AGC circuit 22, and video A / D converter 23 is controlled by a timing signal generated by a timing generator 24. In the timing generator 24, a clock used for signal processing in a data processing encoding / decoding unit 31 (in the video signal processing unit 3) described later is input, and a required timing signal is generated based on this clock. The Thereby, the signal processing timing in the camera block 2 is synchronized with the processing timing in the video signal processing unit 3.

The camera controller 25 performs necessary control so that each functional circuit unit provided in the camera block 2 operates properly, and performs autofocus, automatic exposure adjustment, aperture adjustment, zoom, and the like on the lens block 1. It is supposed to perform control for
For example, in the case of autofocus control, the camera controller 25 controls the rotation angle of the focus motor based on focus control information obtained according to a predetermined autofocus control method. As a result, the imaging lens is driven so as to be in a just-focus state.

  At the time of recording, the video signal processing unit 3 performs compression processing on the digital image signal supplied from the camera block 2 and the digital audio signal obtained by collecting the sound with the microphone 202, and converts the compressed data into user recording data. To the media drive unit 4 in the subsequent stage. Further, an image generated from the digital image signal and the character image supplied from the camera block 2 is supplied to the viewfinder drive unit 207 and displayed on the viewfinder 204.

  Further, at the time of reproduction, the user reproduction data (read data from the optical disc 51) supplied from the media drive unit 4, that is, the compressed image data and audio data are demodulated, and these are reproduced as a reproduction image signal and reproduction. Output as an audio signal.

  In the present embodiment, MPEG (Moving Picture Experts Group) 2 is adopted for moving images as a compression / decompression processing method for image data. Further, an ATRAC (Adaptive Transform Acoustic Coding) method (ATRAC, ATRAC2, ATRAC3, etc.) is adopted as the compression / decompression processing method of audio data.

  The data processing encoding / decoding unit 31 of the video signal processing unit 3 mainly includes control processing related to compression / decompression processing of image data and audio data in the video signal processing unit 3 and input of data via the video signal processing unit 3. Execute processing to control output.

  The video controller 33 executes control processing for the entire video signal processing unit 3 including the data processing encoding / decoding unit 31. The video controller 33 includes, for example, a microcomputer, and can communicate with the camera controller 25 of the camera block 2 and a driver controller 46 of the media drive unit 4 described later via a bus line (not shown), for example. Has been.

For example, the sound collected by the microphone 202 is input to the audio compression encoder / decoder unit 32 as digital audio data via the A / D converter 64 (in the display / image / audio input / output unit 6).
The audio compression encoder / decoder unit 32 performs a compression process on the audio data input according to the ATRAC format.

  In the display / image / audio input / output unit 6, the image data input to the video D / A converter 61 is converted into an analog image signal here and branched to the display controller 62 and the composite signal processing circuit 63. Is output.

  The display controller 62 drives the display unit 6A based on the input image signal. As a result, the reproduced image is displayed on the display unit 6A. Further, in the display unit 6A, not only the display of the image obtained by reproducing from the optical disc 51, but of course, the captured image obtained by photographing with the camera part composed of the lens block 1 and the camera block 2, It is possible to display and output in almost real time.

  In addition to the reproduced image and the captured image, as described above, a message display using characters, characters, or the like for notifying the user of a required message according to the operation of the device is also performed. Such message display should be output from the data processing encode / decode unit 31 to the video D / A converter 61 so that, for example, a required character or character is displayed at a predetermined position under the control of the video controller 33. What is necessary is just to perform the process which synthesize | combines image data, such as a required character and a character, with respect to image data.

  The composite signal processing circuit 63 converts the analog image signal supplied from the video D / A converter 61 into a composite signal and outputs it to the video output terminal T1. For example, if an external monitor device or the like is connected via the video output terminal T1, an image reproduced by the video camera can be displayed on the external monitor device.

  Further, in the display / image / audio input / output unit 6, the audio data input from the audio compression encoder / decoder unit 32 to the D / A converter 65 is converted into an analog audio signal here, and is supplied to the headphone / line terminal T2. Is output. Further, the analog audio signal output from the D / A converter 65 is also branched and output to the speaker SP via the amplifier 66, whereby reproduced audio or the like is output from the speaker SP. Become.

  In the media drive unit 4, the recording data is encoded so as to be suitable for disc recording and transmitted to the deck unit 5 during recording, and the data read from the optical disc 51 in the deck unit 5 is decoded during reproduction. To obtain reproduced data and transmit it to the video signal processing unit 3.

  The encoder / decoder 41 of the media drive unit 4 receives recording data (compressed image data + compressed audio data) from the data processing encoding / decoding unit 31 at the time of recording. The encoded data is stored in the temporary buffer memory 42. Then, the data is transmitted to the deck unit 5 while being read out at a required timing.

  At the time of reproduction, the digital reproduction signal read from the optical disc 51 and inputted through the RF signal processing circuit 44 and the binarization circuit 43 is subjected to decoding processing according to a predetermined format, and video is reproduced as reproduction data. The data is transmitted to the data processing encoding / decoding unit 31 of the signal processing unit 3.

Also in this case, if necessary, the reproduction data is temporarily stored in the buffer memory 42, and the data read therefrom at a required timing is transmitted and output to the data processing encoding / decoding unit 31. Such write / read control for the buffer memory 42 is executed by the driver controller 46.
It should be noted that, for example, during reproduction of the optical disk 51, even if the servo is disconnected due to disturbance or the like and reading of the signal from the disk becomes impossible, the read data is stored in the buffer memory 42 within the period. If the playback operation on the disc is restored, it is possible to maintain time-series continuity as the playback data.

The RF signal processing circuit 44 performs necessary processing on the read signal from the optical disc 51, such as an RF signal as reproduction data, a focus error signal for servo control for the deck unit 5, and a tracking error signal. Servo control signal is generated. The RF signal is binarized by the binarization circuit 43 as described above, and is input to the encoder / decoder 41 as digital signal data.
The generated servo control signals are supplied to the servo circuit 45. The servo circuit 45 executes necessary servo control in the deck unit 5 based on the input servo control signal.

  The deck unit 5 is a part composed of a mechanism for driving the optical disc 51. Although not shown here, the deck unit 5 has a mechanism in which the optical disk 51 to be loaded is detachable and can be exchanged by the user's work.

In the deck unit 5, the loaded optical disk 51 is rotationally driven by a spindle motor 52. The optical disk 51 is irradiated with laser light from the optical head 53 during recording / reproduction.
The optical head 53 performs high-level laser output for heating the recording track to the Curie temperature during recording, and relatively low-level laser output for detecting data from reflected light by the magnetic Kerr effect during reproduction. . Therefore, although not shown in detail here, the optical head 53 is equipped with a laser diode as a laser output means, an optical system including a polarizing beam splitter, an objective lens, and the like, and a detector for detecting reflected light. Yes. The objective lens provided in the optical head 53 is held so as to be displaceable in a disk radial direction and a direction in which it is in contact with and separated from the disk, for example, by a biaxial mechanism.
Although not shown, the deck unit 5 includes a thread mechanism that is driven by a thread motor 54. By driving this sled mechanism, the entire optical head 53 is movable in the disk radial direction.

  The operation unit 7 is provided with various operators for performing various operations of the video camera. Various kinds of user operation information using these operators are supplied to the video controller 33, for example. The video controller 33 supplies the camera controller 25 and the driver controller 46 with operation information and control information for causing each unit to execute necessary operations according to user operations.

The external interface 8 is provided so that data can be transmitted between the video camera and the external device. For example, as shown in the figure, the external interface 8 is provided between the I / F (interface) terminal T3 and the video signal processing unit. It is done. The external interface 8 is not particularly limited here. For example, IEEE 1394 or the like may be adopted.
For example, when an external digital image device and the video camera of this embodiment are connected via the I / F terminal T3, it is possible to record an image (sound) captured by the video camera on the external digital image device. It becomes. In addition, image (sound) data or the like reproduced by an external digital image device can be recorded via the external interface 8 to be recorded on the optical disc 51 according to the disc format. Furthermore, for example, a file as character information used for inserting a caption can be captured and recorded.

The timer 34 measures the current date and time, and when requested by the video controller 33, the current date and time can be output to the video controller 33.
The flash memory 35 is a kind of so-called non-volatile memory, and can store information on main subjects and photographers input by the user.

  The power supply block 9 supplies a power supply voltage of a required level to each functional circuit unit using a DC power source obtained from a built-in battery or a DC power source generated from a commercial AC power source. The video controller 33 controls the power on / off by the power block 9 according to the operation of the operation unit 7.

2. Example of encoding block configuration

Next, the configuration of the encoding block 31A provided in the data processing encoding / decoding unit 31 shown in FIG. 1 will be described with reference to FIG.
In FIG. 2, a filter unit 71 performs color signal and resolution conversion on the image data from the camera block 2, and displays an image in units of macroblocks composed of 16 × 16 pixel blocks as basic processing units for image compression. The data is output to the subtracter 72. A part of the image data input to the filter unit 71 is supplied to the motion detection unit 78.

  The motion detection unit 78 performs motion detection within a range of several tens to several hundreds of frames in units of macro blocks while using the image memory 80 as a work area based on the control of the memory control unit 79, for example. Then, this detection result is output to the motion compensation unit 81 as motion vector information.

Based on the detection result of the motion vector detection unit 78, the motion compensation unit 81 performs image processing such as motion compensation such as partially moving the encoded image of the preceding and following frames.
The subtracter 72 takes out the difference between the image data from the filter unit 71 and the motion compensation unit 81 and outputs the difference to the DCT unit (discrete cosine transform unit) 73.

  The DCT unit 73 performs DCT conversion on the difference data extracted by the subtracter 72 and then outputs the result to the quantization unit 74.

  In the quantization unit 74, in order to reduce the amount of data, requantization is performed to reduce high frequency components. In this case, the quantization unit 74 determines the quantization coefficient for each macroblock using face image information (object information) from the object detection block 90 described later and verification information from the verification unit 101 described later. Is done. That is, the compression rate for each macroblock is determined.

The output of the quantization unit 74 is supplied to the scan unit 75. In the scanning unit 75, for example, the block order is rearranged so as to be zigzag from the block at the upper left of the screen of one frame, and is supplied to the variable length coding unit 76. The variable length encoding unit 76 performs run length coding to further compress the information amount.
The encoded data output from the variable length encoding unit 76 is supplied to the multiplexer 77, where the audio data from the audio compression encoder / decoder unit 32 is multiplexed and output as a bit stream. .

  The output of the quantization unit 74 is also used as comparison data for taking a difference. For this reason, it is also supplied to the local decoder block constituted by the inverse quantization unit 82 and the inverse DCT unit 83. The output of the inverse DCT unit 83 and the output of the motion compensation unit 81 are added by an adder 84 and are held in the image memory 86 via the memory control unit 79.

3. Object detection block configuration

Next, the object detection block shown in FIG. 2 will be described with reference to FIGS. In the present embodiment, description will be made assuming that the object detection block 90 detects a human face image as an object.
FIG. 3 is a diagram showing the configuration of the object detection block. FIG. 4 is a diagram showing an example of a face image recognition sequence in the object detection block.

  The object detection block 90 shown in FIG. 3 includes an object detection unit 91, an object parameter detection unit 94, and a wavelet conversion unit 99.

The object detection unit 91 is, for example, a block that detects a rough face position from input image data. In this case, the object detection unit 91 includes a motion detection unit (Head Detector) 92 and a face detection block (Face Finder) 93. .
The motion detection unit 92 detects a rough face position in which a face image is included in the input image data by detecting a portion having a motion from the image data of the previous and subsequent frames of the input image data.
A face image detection unit (Face Finder) 93 sets feature points called nodes on the facial organs for the input image data, and detects image portions that appear to be faces by graph matching. To be done.

  As the image data input to the motion detection unit 92 and the face image detection unit 93 of the object detection unit 91, the image data 80 includes a number in the motion vector detection unit 78 shown in FIG. The image data is stored in units of frames. That is, the object detection block 90 performs object detection using frame image data stored in the image memory 80.

  When the object detection block 90 cannot detect the face image in real time, the object detection block 90 reads the image from the image memory 80 so that the face detection is performed discretely in the object detection block 90. What is necessary is just to select the period which takes in data.

  The object parameter detection unit 94 is a block that detects and processes the detailed data of the face image detected by the object detection unit 91, and includes a face detection block (Pre Selector) 95, a face enlargement / reduction unit 96, a face image. An area detecting unit (Landmark Finder) 97 and a background erasing unit 98 are included.

The face image detection unit 95 receives an input image in units of frames in which the face position is roughly detected by the object detection unit 91, for example, an input image as shown in FIG.
The face image detection unit 95 assigns about 50 nodes (feature points) to the input image shown in FIG. 4A, for example, as shown in FIG. Detection is performed, and a feature amount is detected from a directionality, shading, positional relationship, and the like in each node by a wavelet transform unit 99 described later.

  The image detected by the face image detection unit 95 is supplied to the face enlargement / reduction unit 96, and the face enlargement / reduction unit 96 enlarges / reduces the image to an optimal size, and then The area detection unit 97 cuts out the face area by following each node and meeting the graph at an appropriate position.

  In the background erasing unit 98, a face image other than the face region cut out by the face region detecting unit 97 is output as a detected image, such as a face image filled with gray. As a result, an extracted image obtained by extracting only the face portion as shown in FIG. 4C is output as face image information for recognizing the face image.

  For example, when a feature point called a node is set on a facial organ in a face image, the Wavelet conversion unit 99 detects a feature amount such as directionality, shading, and positional relationship at each node.

  Note that the object detection block 90 cuts out a face image area from the input image even when the input image includes a plurality of face images as shown in FIG. It can be output as information.

  As described above, in the video camera according to the present embodiment, the object detection block 90 detects whether or not a face image is included as an object in the image captured by the camera block 2. When an image / face image is included in the input image from the camera block 2, information about the face image is output to the quantization unit 74 shown in FIG. 2 as object information.

In the quantization unit 74, when the image data input from the DCT unit 73 is quantized based on the object information input from the object detection block 90, for example, the quantization coefficient of the macroblock including the face image is set. It is small (compression ratio is small). For other macroblocks, the quantization coefficient is set larger (the compression ratio is larger) than the macroblock including the face image.
Note that the value of the quantization coefficient for quantizing the macroblock including the face image may be appropriately set in consideration of the image quality. The quantization coefficient values in other regions may be set in consideration of the total information amount.

Therefore, with this configuration, the image data of the macroblock including the face image can be quantized while maintaining a small quantization coefficient in the quantization unit 74, and the face image regarded as important The portion can have high image quality.
For image data other than macroblocks including face images, the amount of data for image data other than macroblocks including face images can be reduced by increasing the quantization coefficient. Thereby, for example, the amount of image information can be reduced as compared with the case where the entire frame image is compressed with a small quantization coefficient.

4). Other configuration example of encoding block

Next, another configuration of the encoding block 31A of the data processing encoding / decoding unit 31 shown in FIG. 1 will be described with reference to FIG. In addition, illustration of the part 32 is abbreviate | omitted in the same site | part as the said FIG.2 and FIG.3. In FIG. 6, the audio compression encoder / decoder 32 is not shown.

  In the encoding block 31A of the data processing encoding / decoding unit 31 shown in FIG. 6, a database 100 serving as storage means for registering a face image considered important in advance, and the database 100 are registered in the database 100. A collation unit 101 that collates the face image with the face image detected from the input image data in the object detection block 90 is provided.

For example, the database 100 may be configured using the optical disc 51, or may be configured using a memory provided in the verification unit 101.
When the database 100 is configured using the optical disc 51, the optical disc 51 normally stores a data area 110 for storing image content, address information of the data area 110, content attributes, and the like, as shown in FIG. The management information area 111 is divided into
Therefore, it is conceivable to store face image data or the like as a registered image using the management information area 111 of the optical disc 51.
In this case, for example, when the disc 51 is loaded, the registered image is read from the optical disc 51 and the registered image is taken into the memory in the collation unit 101.

  For example, at the time of recording, the registered image taken into the memory in the collation unit 101 is read out, and collation with the face image information extracted by the object detection block 90 is performed in the collation unit 101. It should be noted that such a configuration of the database 100 is merely an example, and it goes without saying that other configurations may be used.

  In the collation unit 101, for example, information obtained from each node of the detected image as shown in FIG. 8A and information on the registered image registered in advance in the database 100 as shown in FIG. 8B. Compare and match. The collation result of the collation unit 101 is supplied to the quantization unit 74. In this case, the collation unit 101 collates the face image information from the object detection block 90 and the face image information in the database 100 at a cycle in which the object detection block 90 fetches image data from the image memory 80. Just do it.

  When the quantization unit 74 quantizes the image data input from the DCT unit 73 based on the collation result from the collation unit 101, for example, it matches the face image registered in the database 100 in the collation unit 101. The quantization coefficient is reduced for the frame-based image data including the face image, and the quantization coefficient is increased for the other image data.

Therefore, even in such a configuration, for example, for the frame image data in which a face image registered in advance as a highly important image in the database 100 is included, the quantization unit 74 performs a small quantum for each frame image. Since quantization is performed by the quantization coefficient, high quality of the face image can be achieved.
Further, since the frame image data not including the face image is quantized with a larger quantization coefficient than the portion including the face image, the amount of image information of the entire frame image can be reduced.

Further, for example, if the detection result of the object detection block 90 is supplied to the quantization unit 74 together with the matching result from the matching unit 101 as indicated by a broken line, a specific face image is included in a specific image. It is possible to reduce only the quantization coefficient of the face image portion to improve the image quality, and for other images, the quantization coefficient can be increased to reduce the amount of image information.
Therefore, in the case of such a configuration, only the information amount of the face image portion that matches the face image registered in the database 100 among the data from the DCT unit 73 increases, so that the image information amount of the entire frame image Can be greatly reduced.

  If the collation result of the collation unit 101 is supplied to the driver controller 46 of the media drive unit 4 shown in FIG. 1, whether or not the driver controller 46 records the bit stream from the video signal processing unit 3 on the optical disc 51. It is also possible to control.

  For example, a face image is detected by the object detection block 90 from the images shown in FIGS. 9A, 9B, and 9C, and among the face images, the face image shown in FIG. Assume that a collation result is obtained that matches the face image registered in the database 100.

  Then, by supplying such a collation result to the driver controller 46 of the media drive unit 4 shown in FIG. 1, the driver controller 46 uses the optical disc 51 for data corresponding to the images shown in FIGS. It is possible to perform control so that only the data corresponding to the image shown in FIG.

In this case, only an image including a face image that matches the face image registered in the database 100 is recorded on the optical disc 51.
Thereby, only the frame image data containing the important face image is recorded on the optical disc 51, and the recording capacity of the optical disc 51 can be used effectively.

  Further, after the collation result of the collation unit 101 is supplied to the driver controller 46 of the media drive unit 4 shown in FIG. 1, the quantization unit 74 identifies the specific face image from the frame image as described above. If the image shown in FIG. 9B is recorded on the optical disc 51, only the image of the face portion is improved in image quality. Can be recorded.

  In this way, on the optical disc 51, only an image that includes a face image that matches the face image registered in the database 100 and that is a high-quality image of the face image can be recorded. Therefore, the amount of image information can be further reduced, and the recording capacity of the optical disc 51 can be used more effectively.

When the database 100 is configured using the optical disc 51 as in the present embodiment, for example, the optical disc 51 includes a face image that matches the registered image (face image) registered in the database 100. Only the input image that has been recorded can be recorded.
Therefore, for example, if a face image that the user wants to record is registered in the database 100 of the optical disc 51, it becomes possible to create the optical disc 51 in which only such a face image is recorded.

5). Decoding block configuration example

In the above description, the operation in the case of encoding in the data processing encoding / decoding unit 31 in order to record the captured image taken by the camera block 2 on the optical disc 51 has been described as an example. The present invention can be applied even when decoding is performed in the decoding unit.

FIG. 10 is a diagram illustrating a configuration example of the decode block 31B of the data processing encode / decode unit 31.
In the decoder block 31B shown in FIG. 10, description of audio data read from the optical disc 51 is omitted, and only video data is described.
Further, the decoding block shown in FIG. 10 includes two frame memories 126 and 127, and these frame memories 126 and 127 are alternately used for an I picture and a P picture.

  A bit stream (video bit string) read from the optical disc 51 of the deck unit 5 and obtained by performing decoding processing according to a predetermined format in the media drive unit 4 is sent to the variable length code decoder via the reception buffer 121. 122 is input. Then, the variable length code decoder 122 performs a decoding process on the variable length encoding of the encoding block 31A described above.

In the variable-length code decoder 122, a picture type Ptype indicating whether the picture is an I picture, a P picture, or a B picture, intra prediction (intra), backward / forward prediction (B / F), forward / backward prediction (F / B) is separated into a macroblock type Mtype or the like.
The picture type Ptype separated by the variable length code decoder 122 is used for switching control of the selector switch 131, and the macroblock type Mtype is used for switching control of the selector switch 132.
The motion vector MV separated by the variable length code decoder 122 is used in the half-pel motion compensation units 128 and 129 from the frame memories 126 and 127.

Here, the picture type Ptype will be briefly described. For each frame of the image signal to be compressed, three types of image data (one frame of video data) having different compression degrees are provided. These are called an I picture (Intra Picture), a P picture (Predicted Picture), and a B picture (Bidirectionally predicted Picture).
The I picture is image data including only the intra-frame prediction screen, and the P picture is image data generated by inter-frame forward prediction. The B picture is image data generated by inter-frame bi-directional prediction.

Also, the quantized DCT coefficient (two-dimensional variable length code) decoded by the variable length code decoder 122 is converted by the inverse quantization unit 123 and the inverse DCT conversion unit (IDCT) 124 to become difference image data. If the picture is an I picture or a P picture, the adder 125 adds the pixel values of the corresponding positions, which are motion-compensated for each macroblock, in one of the frame memories 126 and 127, and The frame memory 127 is stored.
During the decoding process of the I picture and P picture, one of the frame memories 126 and 127 that are not currently being written is output.

  In the B picture, half-pel motion compensation is performed in parallel from the two frame memories 126 and 127 by the half-pel motion compensation units 128 and 129 that perform motion compensation in units of half pixels. The outputs of the half-pel motion compensation units 128 and 129 are averaged by the averaging circuit unit 130, and the motion compensated predicted pixel values are added by the adder 125 using the motion vectors prepared for each of them. A playback image of a B picture is created. This image is output as it is without accumulating anywhere.

  As described above, in the decoder block shown in FIG. 10, an image being decoded is output as it is for a B picture, and a reference image used for half-pel motion compensation is output for an I picture or a P picture. Therefore, the order of the screen after decoding (delay of I picture and P picture) is performed.

  When the present invention is applied to such a decoder block, for example, an I picture and a P picture stored in the frame memories 126 and 127 are supplied to the object detection block 90 while scanning at appropriate intervals. Yes.

The object detection block 90 in this case also has the same configuration as described above, detects a face image from input image information of an I picture and a P picture, and according to the detection result, the driver controller 46 of the media drive unit 4. The face image information is output to the screen.
The driver controller 46 controls the data reading operation from the optical disc 51 based on the face image information from the object detection block 90.

  Accordingly, for example, high-speed picture search in which the media drive unit 4 reads data from the optical disk 51 of the deck unit 5 at high speed until the face image is detected in the object detection block 90 can be realized.

  Further, the collation unit 101 collates the face image detected in the object detection block 90 with a database in which face images registered in advance in the database 100 are registered, and only an image including a face image that matches the face image in the database 100 is included. It is also possible to search and play back.

By the way, in the present embodiment described so far, when specifying a person, the face image information detected in the object detection block 90 and the face image information registered in the database 100 are collated to specify the person. I have to.
However, even if the person is specified in this way, an error may occur in the collation result when the face of the person is extremely similar.

  Therefore, for example, it is conceivable to perform speaker recognition using voice data in combination with the collation of face image information in the object detection block 90 as described above to specify a person more reliably.

6). Configuration example of speaker recognition unit

Here, FIG. 11 shows an example of the speaker recognition block.
Such a speaker recognition block is provided in, for example, the voice compression encoder / decoder unit 32 shown in FIG.
The feature parameter extraction unit 141 shown in FIG. 11 extracts feature parameters from the input voice data and supplies them to the similarity calculation unit 142.
The similarity calculation unit 142 is supplied with the feature parameters extracted by the feature parameter extraction unit 144 from voice data registered in advance in the database 100 or the like, and the similarity calculation unit 142 calculates the similarity from these parameters. To be calculated.
Based on the calculation result of the similarity calculation unit 142, the speaker recognition determination unit 143 determines whether the speaker is a corresponding speaker, and the determination result is output as speaker information.
Accordingly, if this speaker information is supplied to the quantization unit 74 of the data processing encoding / decoding unit 31, the quantization unit 74 recognizes a face image for recognizing a face image from the speaker information and the object detection block 90. It becomes possible to specify a person (object) or the like together with image information (object information).

  If the person is specified by using the face image information detected by the object detection block 90 and the motion vector together, and further the person is specified including the speaker recognition information described above, the person is specified. It becomes possible to carry out more reliably.

  In the present embodiment, the object detection block 90 has been described with reference to an example in which a face image is detected as an object. However, this is merely an example, and other objects such as automobiles are used in the object detection block. It may be configured to detect the shape or character.

In the present embodiment, the case where the present invention is applied to a video camera has been described as an example. However, this is merely an example, and the present invention can be applied to various recording / playback apparatuses other than a video camera. .
Further, as the optical disc 51 that can be recorded as a recording medium, various discs such as a CD (Compact Disc) disc, a DVD (Digital Versatile Disc) disc, and a Blu-Ray disc are described as examples. However, this is merely an example, and a mini disk, a HDD (hard disk drive), a memory card such as a flash memory, or the like can be used as a recording medium.

It is the block diagram which showed the structure of the video camera of embodiment of this invention. It is the figure which showed the structure of the data processing encoding part. FIG. 3 is a diagram illustrating a configuration example of an object detection block illustrated in FIG. 2. It is the figure which showed an example of the sequence of face image recognition. It is the figure which showed an example of the detection of the face image by face image recognition. It is the figure which showed the other structure of the data processing encoding part. It is the figure which showed an example of the database. It is a figure for demonstrating the collation in a collation part. It is a figure for demonstrating the recording operation at the time of recording. It is the figure which showed the structure of the data processing decoding part. It is the block diagram which showed the structure of speaker recognition.

Explanation of symbols

  1 lens block, 2 camera block, 3 video signal processing unit, 4 media drive unit, 5 deck unit, 31 data processing encode / decode unit, 32 audio compression encoder / decoder unit, 33 video controller, 46 driver controller, 51 optical disc, 71 Filter unit, 72 Subtractor, 73 DCT unit, 74 Quantization unit, 75 Scan unit, 76 Variable length encoding unit, 77 Multiplexer, 78 Motion vector detection unit, 79 85 Memory control unit, 80 86 Image memory, 81 Motion compensation unit, 82 inverse quantization unit, 83 inverse DCT unit, 84 adder, 90 object detection block, 91 object detection unit, 92 motion detection unit, 93 face image detection unit, 94 parameter detection unit, 95 face image detection unit , 96 Enlarging / reducing part, 97 Image area detection unit, 98 background elimination unit, 99 Wavelet conversion unit, 100 database, 101 verification unit, 110 data area, 111 management information area, 121 reception buffer, 122 variable length code decoder, 123 inverse quantization unit, 124 ICDT, 125 adder, 126 127 frame memory, 128 129 half-pel motion compensation unit, 130 averaging unit, 141 feature parameter extraction unit, 142 similarity calculation unit, 143 speaker recognition determination unit, 144 feature parameter extraction unit

Claims (9)

Detect specific objects included in the input image information,
When the specific object is detected, at least the specific object included in the input image information is compressed by changing a compression rate. Object detection means for detecting a specific object included in the input image information;
Image compression means for changing the compression ratio of at least the specific object included in the input image information when the specific object is detected by the object detection means;
A signal processing apparatus comprising: The image compression means includes
When the specific object is detected by the object detection means, the image information of the area including the specific object in the input image information is compressed at a compression rate smaller than the image information of the other areas. The signal processing apparatus according to claim 2. Storage means for storing information on the required object;
Collation means for collating whether or not the specific object detected by the object detection means matches the required object stored in the storage means;
The image compression means includes
The input image that includes the specific object when the collating unit determines that the specific object detected by the object detecting unit matches the required object stored in the storage unit The signal processing apparatus according to claim 2, wherein the compression is performed by changing a compression rate of the entire information. The image compression means includes
The input image that includes the specific object when the collating unit determines that the specific object detected by the object detecting unit matches the required object stored in the storage unit The signal processing apparatus according to claim 4, wherein image information in an area including the specific object is compressed at a compression rate smaller than that of image information in other areas. Object detection means for detecting a specific object included in the input image information;
Storage means for storing information on the required object;
Collation means for collating whether or not the specific object detected by the object detection means matches the required object stored in the storage means;
The specific object is included when it is determined that the specific object detected by the object detection means matches the required object stored in the storage means based on the collation result of the collation means. Recording control means for performing recording control for recording the input image information on a recording medium;
A recording apparatus comprising: When the specific object is detected by the object detecting means, the image information of the region including the specific object is included in the input image information including the specific object, and the image information of the other region is included. The recording apparatus according to claim 6, further comprising image compression means for compressing at a smaller compression rate. Reading means for reading image information from the recording medium;
Decoding means for performing predetermined decoding processing on the image information read by the reading means and outputting as image information;
Object detection means for detecting a specific object from the image information output from the decoding means;
Read control means for controlling the reading operation of the image information from the recording medium by the read means based on the detection result of the object detection means;
A playback device comprising: Storage means for storing required objects;
Collation means for collating whether or not the specific object detected by the object detection means matches the required object stored in the storage means;
The read control means includes
9. The playback apparatus according to claim 8, wherein the reading operation of the image information from the recording medium by the reading unit is controlled based on a collation result of the collating unit.

Images