JP2000069348A - Image pickup device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent malfunctions due to surrounding sound or the like at voice input. SOLUTION: First, a proper value is set as a threshold for a coincidence rate in this image pickup device (step S1). The pattern of a frequency spectrum (input pattern) is generated from an input voice (S2). Coincidence rate is calculated, based on the input pattern and a registration pattern of various control commands (S3). If there is no pattern whose coincidence rate exceeds a threshold (S4), when a command providing a high coincidence rate indicates video recording start or stop (S5), a higher threshold is set and the coincidence rate and the threshold are compared again. As a result, if the coincidence rate is lower than the threshold set higher (S7), a command with the highest coincidence rate is outputted (S8). When the command with a high coincidence rate neither indicates video recording start nor video recording stop (S5), a command with the highest coincidence rate is outputted (S9), and the processing returns to the step S1.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an imaging device,
More specifically, the present invention relates to a video camera that recognizes the voice of a photographer and controls various functions according to the recognition result.

[0002]

2. Description of the Related Art In recent years, while video cameras have become very multifunctional, the size of the main body has been reduced. Along with that, many operating members such as push buttons, slide switches and rotary levers are used to operate these functions,
The operation method becomes difficult to understand, and many operation members are arranged in a limited size of the video camera, which impairs operability.

[0003] As means for solving such problems,
There was a method using a hierarchical menu mechanism. this is,
A function item and its setting state are displayed on a display means such as a viewfinder, and an operator sets various functions while watching the display. For example, with at least three keys, a menu on / off key, a setting item selection key, and a setting state change key, about ten or more types of functions can be set.

[0004]

However, the hierarchical menu structure is not difficult to use in the case of a very simple one, but when the hierarchy becomes deeper or the number of items increases,
The operation of traversing the hierarchy has become necessary, which has been cumbersome and complicated for the photographer or user. In addition, there is a problem that quick operation is difficult and lacks immediacy.

[0005] An object of the present invention is to provide an image pickup apparatus that solves such inconveniences.

Another object of the present invention is to provide an image pickup apparatus capable of setting various functions by voice.

[0007]

According to the present invention, there is provided an imaging apparatus comprising: a voice input unit for inputting a voice of a user; a voice pattern holding unit for storing voice pattern information of a content corresponding to a voice command; A voice recognition unit that matches the voice input by the input unit with a voice pattern held in the voice pattern holding unit to recognize voice content, and outputs a control signal corresponding to the recognition result; and An image pickup apparatus comprising: a control unit that controls each unit according to an output control signal, wherein the voice identification unit determines a reliability of a match between an input voice and a voice pattern held in the voice pattern holding unit. It is provided with a reliability determination means for determining, and the threshold value of the reliability is different depending on the control target.

[0008]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a schematic block diagram showing an embodiment of the present invention. Reference numeral 10 denotes a photographic lens, 12 denotes a lens drive motor for moving the photographic lens in the optical axis direction, 14 denotes an image sensor that converts an optical image by the photographic lens 10 into an electric signal, and 16 denotes a light amount incident on the image sensor A CDS 18 samples and holds an image signal output from the image sensor 14 and outputs the signal at an appropriate signal level.
The / AGC circuit 20 is a digital signal processing circuit that converts an analog output of the CDS / AGC circuit 14 into a digital signal and performs digital signal processing.

Reference numeral 22 denotes an image compression circuit for compressing a digital image signal output from the digital signal processing circuit 20, 24 denotes a recording medium for recording the compressed image information output from the image compression circuit 22, and 26 denotes a recording medium. An image decompression circuit which decompresses the compressed image information reproduced from 24 and returns it to the original digital image signal.

Reference numeral 28 denotes a switch for selecting the output of the digital signal processing circuit 20 (ie, the output image of the image sensor 14) or the output image of the image expansion circuit 26 (reproduced image).
Converts the image signal selected by the switch 28 into NTSC
NTS that converts to the format and outputs to the outside from output terminal 32
The C encoder 34 is a liquid crystal drive circuit that drives the liquid crystal display 36 to display an image selected by the switch 28.

38 is a system control circuit for controlling the whole,
Reference numeral 40 denotes operation keys for inputting various instructions to the system control circuit 38. However, in this embodiment, although the details will be described later, a certain operation can be input to the system control circuit 38 without using the operation keys. Reference numeral 42 denotes the focus of the photographing lens 10 and the aperture 1 under the control of the system control circuit 38.
6 and a camera control circuit that controls the image sensor 14.

Reference numeral 44 denotes a microphone for capturing a user's voice;
46 is an A / D converter for digitizing the analog audio output of the microphone 44, 48 is an AGC circuit for automatically adjusting the audio level of the audio data output from the A / D converter 46, 50 is an AGC circuit 48 Is a speech recognition circuit for recognizing the output speech data. When recognizing a command to the system control circuit 38, the voice recognition circuit 50 outputs a control command according to the content of the command to the system control circuit 38.

The user's voice is converted by a microphone 44 into an electric audio signal. And the A / D converter 46
Converts the output of the microphone 44 into a digital signal after passing through an LPF for band control for preventing aliasing noise due to sampling. The AGC circuit 48 adjusts the audio data output from the A / D converter 46 to a predetermined level and supplies the audio data to the audio recognition circuit 50.

FIG. 2 is a schematic block diagram of the speech recognition circuit 50. 52 is a band-pass filter (BPF) for extracting a voice band from the output signal of the AGC circuit 48;
4 is a fast Fourier transform circuit (FFT) for analyzing the frequency spectrum by Fourier transforming the output of the BPF 52;
Is a level comparison circuit that compares the frequency spectrum obtained by the FFT 54 with a predetermined threshold value Vth. 58 is a frequency spectrum pattern (hereinafter, referred to as a registered pattern) 60 of a photographer's voice registered in advance and a level comparison circuit 56. This is a discrimination circuit that discriminates the instruction content of the input voice by comparing the result of the comparison with the above and generates a control command according to the discrimination content.

The registration pattern 60 includes a pattern corresponding to a function of the video camera, such as menu setting, recording start / stop and reproduction start / stop.

In such a configuration, the BPF 52
An audio band is extracted from the output signal of the AGC circuit 48, and the signal of the extracted audio band is converted into a frequency spectrum by the FFT 54. The level comparing circuit 56 compares the frequency spectrum output from the FFT 54 with a predetermined threshold value Vth. At this time, thinning out and interpolation of data are performed in order to align the time axis with the registered pattern 60. The level comparison circuit 56 switches a binary signal of “1” or “0” depending on whether each frequency spectrum exceeds the threshold level Vth. FIG. 3 shows a matrix display example of the output of the level comparison circuit 56. FIG. 4 shows a matrix display example of the registration pattern 60. As shown in FIG. 4, in this example, the registration pattern 60 is defined by 1 to 10 in the time direction, and accordingly, the output of the level comparison circuit 56 also becomes It is defined by 1 to 10 in the time direction. Each frequency component is binarized and represented by 0 or 1.

The discrimination circuit 58 compares the matrix shown in FIGS. 3 and 4 for each element, and outputs a corresponding control signal only when the frequency spectrum of the input voice is similar to the registered pattern 60. If the frequency spectrum of the input voice is not similar to any of the registered patterns 60,
The determination circuit 58 regards the input sound as a normal audio signal and does not output any control signal.

The control signal output from the determination circuit 58 is
It is sent to the system control circuit 38 as a command for executing the function of the video camera. If the control pattern is not similar to the comparison pattern, it is regarded as a normal audio signal and no control signal is output.

In this embodiment, the voice recognition circuit 50 compares the matrix of the registered pattern 60 with the matrix of the input sound to determine whether the command or instruction is a voice command and the content of the command. Specifically, the input sound patterns “1” and “0” are compared with the registered patterns 60 “1” and “0”.
Is calculated, and when the match rate is high, it is determined that the input pattern and the registered pattern 60 are similar. When the match rate is low, the input pattern and the registered pattern 60 are similar.
Are determined not to be similar.

FIG. 4 is a flowchart showing the operation of the determination circuit 58. First, an appropriate value, for example, 80% is set as a threshold value of the coincidence rate (S1). A pattern (input pattern) of the frequency spectrum is created from the input sound (S
2). The input pattern and the registered patterns 60 for various control commands are compared, and the coincidence rate is calculated as described above (S3). In the example shown in FIGS. 3 and 4, the ratio of the input pattern "1" and "0" to the registered pattern "1" and "0" is 54/60 = 90%.

It is checked whether or not at least one of the comparison results between the pattern corresponding to each operation of the registered pattern 60 and the input pattern has a coincidence rate exceeding a threshold value (S4). If there is no matching rate exceeding the threshold (S4), the input voice is determined to be a normal voice, and the process returns to S1 without outputting a command. When there is a command whose matching rate exceeds the threshold value (S4), normally, the command having the highest matching rate should be selected. In this embodiment, before that,
It is checked whether the command having a high matching rate is to start or stop recording (S5).

If the command having a high matching rate is to start or stop recording (S5), a higher threshold is set in the discriminating circuit 58, and the matching rate and the threshold are compared again (S5).
6). As a result, if the matching rate is lower than the increased threshold (S7), the process returns to S1 without outputting the command. If the matching rate is lower than the increased threshold (S7), the command with the highest matching rate is output (S8). The command output here is a command to start or stop recording.

If the command having the highest matching rate is neither the recording start nor the recording stop (S5), the command having the highest matching rate is output (S9), and the process returns to S1.

In this embodiment, the coincidence rate is calculated by a matrix pattern based on the spectrum and the time axis.
The reliability of the input voice command may be determined by another method.

[0026]

As can be easily understood from the above description, according to the present invention, it is possible to set relatively loose judgment conditions except for the command relating to the recording operation, thereby performing more immediate voice recognition. Can be done. Furthermore, the command regarding the video recording operation is determined more strictly than other commands, so that it is possible to prevent malfunction due to unintentional or ambient noise.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of an embodiment of the present invention.

FIG. 2 is a schematic block diagram of a speech recognition circuit 50.

FIG. 3 is a data example of a registration pattern 60;

FIG. 4 is an example of data of an input pattern.

FIG. 4 is an operation flowchart of a determination circuit 58;

[Explanation of symbols]

10: Shooting lens 12: Lens drive motor 14: Image sensor 16: Aperture 18: CDS / AGC circuit 20: Digital signal processing circuit 22: Image compression circuit 24: Recording medium 26: Image expansion circuit 28: Switch 30: NTSC encoder 32 : Output terminal 34: liquid crystal drive circuit 36: liquid crystal display 38: system control circuit 40: operation key 42: camera control circuit 44: microphone 46: A / D converter 48: AGC circuit 50: voice recognition circuit 52: band pass filter (BPF) 54: Fast Fourier Transform circuit (FFT) 56: Level comparison circuit 58: Discrimination circuit 60: Frequency spectrum pattern of the photographer's voice registered in advance

Claims (3)

[Claims] 1. A voice input means for inputting a voice of a user, a voice pattern holding means for holding voice pattern information of a content corresponding to a voice command, and a voice input by the voice input means for holding the voice pattern Voice recognition means for recognizing voice content by comparing it with a voice pattern held in the means, and outputting a control signal corresponding to the recognition result; and controlling each section according to the control signal output from the voice recognition means. An image pickup apparatus having control means, wherein the voice identification means includes reliability determining means for determining the reliability of matching between the input voice and the voice pattern held in the voice pattern holding means, An imaging apparatus, wherein a threshold value of a degree differs for a control target. 2. The imaging apparatus according to claim 1, wherein the threshold value of the reliability is higher for a recording operation than for the other. 3. The imaging apparatus according to claim 1, wherein the voice pattern holding unit holds a voice pattern of the user.