JPS6268384A - Picture communicating system - Google Patents

Abstract

PURPOSE:To efficiently transmit the moving picture at the low transmitting speed by recognizing the picture element of respective parts to constitute the transmitting object picture at the transmitting side, extracting and sending the feature quantity of the part picture and synthesizing and displaying it at the receiving side. CONSTITUTION:An object 12 is photographed by a picture input device 1, taken out as the television signal waveform, fetched into a picture accumulating memory 3 through a recognizing processor 2 and an outline extracting operator is operated to the luminance value of the picture element. The changing point discontinuous to the color information of the picture element is searched, the outline is extracted, pattern matching with the outline picture accumulated as a data base 4 is executed and respective parts are recognized. Next, the obtained feature parameter is sent through a transmitting device 5 to a transmission line 6, received by a receiving device 7, in accordance with the received feature parameter to the picture data in a data base 10 at the receiving side, deformation is given to respective feature parts by a reproducing processing device 8, synthesizing is executed, the luminance value and the color value of the picture element are written to a displaying memory 9, a synthetic picture is converted to the television signal and sent to the displaying device.

Description

DETAILED DESCRIPTION OF THE INVENTION (1) Technical field to which the invention pertains The present invention relates to a single-image communication method, in particular, to efficiently process moving images by applying image processing technology to moving images including people and background images. This relates to an image communication method for digitally transmitting images.

(2) Conventional technology In conventional image communication systems, on the transmitting side, the television signal waveform obtained by scanning the image to be transmitted is
The temporal changes in the signal waveform are encoded and transmitted, and the receiving side decodes the code and reproduces the original television signal waveform for communication. Bandwidth compression encoding was used to compress the redundancy of image signals. This band compression encoding method utilizes the spatial correlation between adjacent pixels.

There are methods to reduce the amount of code, and methods to reduce the amount of code by using the temporal correlation of pixels between adjacent frames, but all of them perform encoding using pixel-based processing. However, there was a limit to the compression of the amount of code.

(3) Purpose of the Invention The purpose of the present invention is to extract the characteristic fit of the image by recognizing the image content of the screen represented by the video signal obtained from the television camera, and to calculate its characteristic parameters. By transmitting.

An object of the present invention is to provide an image communication method that efficiently transmits moving images at a very low transmission rate.

(4) Structure of the invention (4-1) Differences between the characteristics of the invention and the conventional technology The present invention has the following features:
The transmitting side recognizes each partial image element that makes up the image to be transmitted, extracts and transmits the feature values of the partial images, and synthesizes and displays them on the receiving side.

At this time, the most important feature is that image recognition and reproduction are performed using an image database on the transmitting side and the receiving side, and the amount of information required for moving image transmission is compressed.

In conventional encoding techniques, one image is encoded pixel-by-pixel, and the required amount of code is proportional to the number of pixels. On the other hand, in the present invention.

Since the image database is prepared corresponding to the constituent elements of one screen, and the relationship indicating the correspondence between the transmission target image and these is encoded, the amount of code can be made regardless of the number of pixels. Furthermore, the amount of database to be prepared can be increased by increasing the code length to the power of the code length, so it is possible to increase the accuracy of the encoded image by increasing the code length to the power of the power.

(4-2) Example The present invention will be described in detail below using two drawings.

FIG. 1 is a diagram illustrating an embodiment of the present invention, in which numeral 1 denotes an image input device such as a television camera, which obtains an image from a subject 12. In FIG. 2 is a recognition processing device that uses the image storage memory 3 and the image database 4 to perform special @
This is a device that recognizes quantities and extracts them as parameters. This parameter is sent out as a signal by the transmitter at 5, and
It is transmitted through the transmission path. The transmitted characteristic parameters are received by a receiving device 7 and sent to a reproduction processing device 8. In this device, an image is synthesized from the received feature parameters using 10 image databases, and is written into a 9 display memory. 11 is a display device, and 12 is a person to whom transmission is performed.

Figure 2 is a diagram showing an example of a person image to be transmitted.
00 is the head of the human figure, 101 is the eyebrow, 102 is the eye, 103
is the nose, 104 is the mouth, and 105 is the chest.

106 indicates an arm.

Figure 3 is a diagram showing an example of how to use the Laplacian operator to extract the contour of an image. (i-1,j), (i,j-
An operator is constructed using four pixels: 1), (t, j+1), and (i+1°j). FIG. 4 is a diagram showing the feature amount of the eye, which is a partial image of the face, where II is the length of the eye, 1
2 is the width of the eye, l is the diameter of the pupil, 14. (1, is a numerical value indicating the position of the pupil.

Figure 5 is a diagram explaining an example of deformation when playing back and displaying the original image, where the solid line is the eye contour before deformation and the pupil position.The dotted line is the eye contour after deformation. indicates the position of the eyes. /,, ff4. l.

is the feature parameter before transformation + '2' + '4Z
β5' is the value of the feature parameter after transformation. Also,
FIG. 6 is a diagram illustrating an example of images stored in the image database, where a represents the head, b represents the eyebrows, C represents the eyes, and d represents the chest.

e is an image of an arm, which is stored separately for each extracted and recognized part.

Next, an example of the image recognition operation of the present invention will be explained using FIGS. 1 to 4. In FIG.

A person 12, which is a target object, is photographed by an image input device 1, extracted as a television signal waveform, and sent to a recognition processing device 2.
The image is taken into the image storage memory 3 via the. At this time, the television signal waveform representing the human image is sampled into pixels corresponding to a square lattice taken on the target screen and stored in the image memory as the luminance value of one pixel and its color value. That is, the human image to be transmitted is stored in the storage memory 3 in the form shown in FIG. 2, with each image component corresponding to a pixel, and its luminance value and color value.

The operation of recognizing partial images of a person's head, eyes, nose, mouth, etc. from this person image and extracting characteristic parts and characteristic parameters will be explained. In order to extract the contour line of the image component from the image taken into the storage memory 3, the contour extraction operator (
For example, the contour line can be extracted by applying a 'Laplacian operator, etc.) and searching for discontinuous change points on the color information of the pixels. As shown in FIG. 3, if the luminance value of each pixel is expressed as f (i, j) for pixel (t, j), then , j) =f(i-1,j)+f(i,j+1)+f(t+1.
j)+f(i, j+1)-4f(i, j) When the difference in luminance between the (t, j) pixel and its surrounding pixels is large, the pixel (i.

j) is a function that gives a large function value, and is an operator that can detect contour lines in an image based on the magnitude of that value. For example, eyebrows, eyes, mouth, etc. may be recognized as characteristic parts from the contour image extracted in this manner by using a pattern matching method as described below.

That is, the face of the target person is photographed in advance as an image, its outline is extracted using the same method as above, and the system constructor extracts the shape of the outline of the partial image of the face corresponding to the eyes, nose, mouth, etc. Recognize and specify.

Each of them is stored as a database in the form shown in Figure 6. Next, pattern matching is performed between the contour line image of the human face extracted as described above and the contour line image of each partial image stored as a database of the layers, eyes, mouth, etc. Each part of the input image, such as the eyebrows, eyes, and mouth, is recognized by calculating the degree of similarity. At this time, the degree to which the shape of each part such as eyebrows, eyes, mouth, etc. is deformed from the image of eyebrows, eyes, mouth, etc. in the database image as a base is determined.

Extract it as a feature parameter of that image. These processing techniques are known as image processing techniques using ordinary computers, and can be easily realized by applying existing image processing devices. Figure 4 shows an example of characteristic parameters obtained from characteristic parts such as eyebrows, eyes, and mouth.The characteristic parameters such as eye length '++ and eye width 12 can be instantly determined by detecting the position of the characteristic parts. can get. The above recognition/extraction processing is performed under the control of a recognition processing device 2 having a configuration similar to that of a typical digital computer. It is clear that recognition of the chest, arms, etc. other than the head of the target image can also be performed in the same manner as described above.

The feature parameters obtained as above are.

It is sent to the transmission path 6 via the transmitting device 5, and is sent to the receiving device 7.
received at

Next, an example of operation for reproducing the original image from the feature parameters received by the receiving device will be described using FIGS. 5 and 6. The receiving side database 10 has the same data input in advance as the sending side database 4.

The image data in the database 10 is transformed into each characteristic part by the reproduction processing device 8 according to the received characteristic parameters. Let's explain this using eye deformation as an example.

The degree of eye opening can be determined from the received feature parameter l,,12, and deformation is applied according to this feature parameter.
The contour line under the eye remains unchanged, and the width of the upper contour line at the center of the eye is jl! 2, a new contour line is generated by approximately interpolating, for example, a circular arc. In this way, you can simulate opening and closing your eyes. Also, for the eyes, the feature parameter + '4
The center position given by +'5 is +! ! 4',
Move to JS'.

By using a circle with a diameter of e, the movement of the pupil can be reproduced. , this transformed image is combined with other characteristic parts, and the brightness values and color values of the pixels are written in the display memory 9. 2 The combined image is converted into a television signal and 9 is sent to the display device. Processing such as database search, transformation, and synthesis is performed under the control of the reproduction processing device 8.

The display device 11 is a device that displays the sent television signal as an image, and an existing image output device may be used.

In order to operate as explained above, the information required for transmission is only the characteristic parameters, and as is clear from this result, moving images can be efficiently transmitted at a much lower transmission speed than conventional technology. There has been an improvement in the ability to

(5) As described in detail, according to the present invention, the transmitting side extracts and transmits the parameters of features and quantities by recognizing the image, and the receiving side extracts the original parameters from the transmitted parameters. Images are synthesized and displayed, and the information required for transmission is only the feature amount, so it is possible to compress the amount of information with high efficiency. 3 It is possible to transmit moving images at a very low transmission speed. becomes possible.

[Brief explanation of drawings]

Fig. 1 shows the configuration of an embodiment of the present invention, Fig. 2 shows an example of a human image to be transmitted, Fig. 3 is a diagram explaining operators used in contour extraction, and Fig. 4 shows features of eyes. FIG. 5 is a diagram for explaining the parameters, FIG. 5 is a diagram for explaining the deformation of the eye, and FIG. 6 is a diagram for explaining an example of images stored in the database. 1... Image input device, 2... Recognition processing device, 3...
・Image storage memory, 4...Image database, 5...
- Transmitting device, 6... Transmission path, 7... Receiving device, 8.
. . . Reproduction processing device, 9 . . . Display memory, 10 . . . Image database. 11... Display device, 12... Person image, 100...
Head of a human figure, 101... Eyebrows, 102... Eyes, 10
3...Nose. 104...mouth, 105...chest, 106...arm, a
···head. b...eyebrows, C...eyes, d...chest, e...arms. Patent applicant: Nippon Telegraph and Telephone Corporation, patent attorney
Mori 1) Kan EZIE (') engraving (no changes to the content) Figure 1 Figure 6 Figure 6 Procedural author's manuscript (method) December 13, 1985 Patent Application No. 208164 2, 1985 Title image of the invention Communication method 3, relationship with the case of the person making the amendment Patent applicant address: 1-6 Uchisaiwai-cho, Chiyoda-ku, Tokyo Name (4)
22) Nippon Telegraph and Telephone Corporation Representative Tsune Shinfuji 4, Agent

Claims (1)

[Claims] In image communication methods that transmit information including image information,
On the transmission side, there is an image input device, a temporary image storage memory, an image database for the image to be transmitted, a recognition processing device that compares the input image and image information in the image database to extract the feature amount of the image to be transmitted, and a transmission device. In addition, the receiving side has a receiving device, an image database, and a reproduction processing device that synthesizes and displays the original image from the received feature amount and information in the image database, and the transmitting side An image communication method characterized in that a feature quantity is sent as a parameter to a receiving side via a transmission path, and the receiving side uses the parameter to reproduce an original image.