JP2919236B2 - Image coding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding apparatus for encoding and transmitting an image signal, and more particularly, to an image encoding apparatus for encoding and transmitting an image signal.
The present invention relates to an image encoding apparatus that extracts a face area from an image signal, sets an area around the face area, and performs a filtering process corresponding to the area, thereby improving the image quality of the face portion of the image.

[0002]

2. Description of the Related Art In recent years, with the development and spread of digital transmission lines, advances in image processing technology, and the development of high-speed digital signal processing technology, demands for realizing an image communication service as a new communication service have been greatly increased. Typical examples of the image communication service include a video telephone service and a video conference service, and these services are becoming actual due to the realization of a high-performance network such as an integrated services digital network (ISDN). . Further, a new communication network represented by a broadband integrated digital network (B-ISDN) has been actively studied for the purpose of providing advanced services.

[0003] In fields other than communications, it is expected that image information (image media) will be handled efficiently. In general, when transmitting an image, the amount of information of the image is enormous, but the amount of information that can be transmitted is limited in terms of the line speed and cost used for transmission. The data is transmitted in a state of being compressed and encoded.

[0004] Studies on image coding have been actively pursued, particularly in the field of communications, as objects of image signal transmission, and various methods have been proposed. Among these image codings, one that has high coding efficiency and has recently been actively used is a discrete cosine transform (DCT: Disc).
An orthogonal transform coding method using a rete cosine transform is known.

[0005] As a known document describing the prior art according to the present invention, for example, there is an "image coding apparatus" in JP-A-4-219089. The publication discloses an input image of a moving image in order to improve the quality of a decoded image in a specific area without causing large distortion in an area other than the specific area such as a face area. A face area detection circuit that detects a face area in the signal and outputs an area specifying signal for specifying the face specific area and the other areas in a distinguished manner, and according to the area specifying signal,
It has a low-pass filter for selectively filtering image signals in an area other than the face area, and a DCT circuit for encoding an output image signal from the low-pass filter.

FIG. 10 is a block diagram of a conventional two-dimensional orthogonal transform coding apparatus. In the figure, 11 is a frame memory unit, 12 is a discrete cosine transform unit, 13 is a quantization unit, and 14 is a coding unit. . An input image signal obtained by digitizing an image signal from a television camera or the like is stored in the frame memory unit 11 for one screen. The discrete cosine transform unit 12 is divided into blocks of N × M pixels (N, M: natural numbers), and performs a two-dimensional orthogonal transform on the pixels of each block. Quantization unit 1
Reference numeral 3 quantizes the transform coefficient obtained by the discrete cosine transform unit 12. The encoding unit 14 encodes the data coefficient quantized by the quantization unit 13.

[0007]

Generally, in a videophone call, a videoconference or the like, the transmitted image often includes a face image of a person, and it is important how the face is captured. However, in the above-described conventional example, regardless of the attribute of the data of the input block, the block simply depends on whether the block is a luminance block or a chrominance block, or the order of coefficients and the amount of a transmission buffer. This has the disadvantage that only the quantization step is controlled, that important areas of the image to be transmitted cannot be transmitted as sharper images, and that points of the image to be transmitted are lost.

The present invention has been made in view of such circumstances, and an image coding apparatus capable of improving image quality so that the image quality of a face area, which is an important area of an image to be transmitted, becomes clearer. It is intended to provide.

[0009]

According to the present invention, in order to achieve the above object, an input image signal is divided into a plurality of blocks, and image data of each block is subjected to two-dimensional orthogonal transform coding to obtain an image. In an image encoding apparatus for compressing data, a face area extracting means for extracting a face area from an image signal stored in a frame memory , and a plurality of steps for performing filtering processing around the extracted face area.
An area setting means for setting a region of which for each area set was characterized Rukoto a filtering means for stepwise varying the coefficients of the filter performs filtering <br/> ring processing It is.

[0010]

When encoding an image signal, a face area is extracted from the image signal stored in the frame memory by the face area extracting means, and filtering is performed around the face area extracted by the area setting means. The surrounding area is set, the filtering means performs filtering processing corresponding to each area set in the area setting means, limits the band of the input image signal, increases the coding efficiency of the surrounding area, and steps the resolution of the surrounding area. Thus, the image quality of the face region is relatively improved while suppressing the deterioration of the visual image quality.

[0011]

Embodiments will be described below with reference to the drawings. FIG. 1 is a block diagram for explaining an embodiment of an image encoding apparatus according to the present invention. In the figure, 1 is a frame memory unit, 2 is an orthogonal transform unit, 3 is a quantization unit, and 4 is an encoding unit. , 5
Denotes a face area extraction unit, 6 denotes an area setting unit, and 7 denotes a filtering unit. A feature of the present invention is that a face area extracting unit 5 that extracts a face area from an input image signal, an area setting unit 6 that sets an area for performing a filtering process, and the area setting unit 6
In that it has a filtering unit 7 that performs a filtering process corresponding to the region set in.

First, the face area extracting unit will be described. The face area extraction unit 5 first extracts skin color pixels using the Y, Cr, and Cb components in order to extract the location of the face. Then, as shown in FIG. 2, FIG. 3, and FIG. 4, the Y, Cr, and Cb components are each binarized with a threshold value for extracting a flesh color pixel to obtain a binary image in which a region including a face portion is extracted. Next, as shown in FIG. 5, a face region is extracted by extracting a common region of three binary images.

Next, the area setting section will be described. The area setting unit 6 sets an area for performing filtering on a peripheral area of the extracted face area. Hereinafter, a method of setting an area used in the present invention will be described with reference to FIG. The maximum value of the horizontal width in the face area extracted by the face area extraction unit 5 is defined as w. In the viewing angle θ obtained from the viewing distances x and w, the width obtained when the eccentricity is increased by 2 degrees is α. An area enlarged by the width α with respect to the face area is set (an area in FIG. 6). Similarly, an area enlarged by the width α with respect to the area is set (area in FIG. 6). In this manner, the peripheral area is sequentially set according to the viewing angle θ. In this embodiment, the viewing distance x is generally
Since it is said that the viewing distance of a video conference or a videophone is 5h to 6h, it is set to 6h.

The filtering section 7 performs a filtering process corresponding to the area set by the area setting section 6. The band of the input image is limited by a low-pass filter having a pass band corresponding to each region, and the resolution of the peripheral region is gradually reduced. FIG. 7 shows the low-pass filter used. Hereinafter, an example of the filtering process will be described. The input image signal stored in the frame memory unit 1
Set a three pixel window. Assuming that the 3 × 3 window of the input image signal is the image signal shown in FIG. 8 and the output of the filter is x ₀ , the output x ₀ of the filter is given by the following equation. X ₀ = (1 · A + KB · B + 1 · C + 1 · D + K ² · E + 1 · F + 1 · G + K · H + 1 · I) / (1 + K) ^{2 The} input image signal is sequentially read from the frame memory unit 1 and filtering processing is performed. The image signal after the processing for one screen is completed is output to obtain an output image signal in which the image quality of the face portion is improved.

The relationship between the filter coefficient and the viewing angle θ when performing the filtering process will be described below with reference to FIG.
As shown in FIG. 9, as the viewing angle θ increases, a smaller filter coefficient K is used. For example, as shown in FIG. 6, the coefficient K is a coefficient K ₃ (viewing angle θ ₁ ) in the area, the coefficient K is a coefficient K ₂ (viewing angle θ ₂ ) in the area, and the coefficient K is a coefficient K _{1 in the area.} (Viewing angle θ ₃ ) and perform filtering processing. That is, as the viewing angle θ increases (the outer region), the filter is processed with a narrower pass band width.

The subsequent operation of the image coding apparatus is shown in FIG.
The same processing as in the conventional example is performed and encoded. That is, the input image signal filtered by the filtering unit 7 is stored in the frame memory unit 1 for one screen. In the orthogonal transform unit 2, N × M pixels (N, M: natural numbers)
And two-dimensional orthogonal transformation is performed on the pixels of each block. The quantization unit 3 quantizes the transform coefficients obtained by the orthogonal transform unit 2. The encoding unit 4 encodes the data coefficient quantized by the quantization unit 3.

[0017]

As apparent from the above description, according to the present invention, stepwise setting a plurality of regions around the facial region of the image, the coefficients of stepwise filter for each region variable
By performing filtering processing by the input image
Since the band of the image signal is limited, the image of the face area is relatively
Assignment of coding to images can be increased,
The image quality of the face area and the surrounding area can be expected.
The drop in image quality of the face area
Image quality can be improved. Further, by performing the filtering process, the image signal is smoothed and random noise is removed, so that the efficiency of motion prediction can be improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram for explaining an embodiment of an image encoding device according to the present invention.

FIG. 2 is a diagram illustrating a binary image extracted with a Y component according to the present invention.

FIG. 3 is a diagram for explaining a binary image extracted with a Cr component according to the present invention.

FIG. 4 is a diagram for explaining a binary image extracted with a Cb component according to the present invention.

FIG. 5 is a diagram for explaining a binary image in which a common part of Y, Cr, and Cb is extracted according to the present invention.

FIG. 6 is a diagram for explaining setting of a filtering area in the present invention.

FIG. 7 is a diagram for explaining a low-pass filter according to the present invention.

FIG. 8 is a diagram for explaining an input image signal according to the present invention.

FIG. 9 is a diagram for explaining a filter coefficient K and a viewing angle θ in the present invention.

FIG. 10 is a configuration diagram of a conventional two-dimensional orthogonal transform encoding device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Frame memory part, 2 ... Orthogonal transformation part, 3 ... Quantization part, 4 ... Encoding part, 5 ... Face area extraction part, 6 ... Area setting part, 7 ... Filtering part.

Claims (1)

(57) [Claims] 1. An image coding apparatus for dividing an input image signal into a plurality of blocks, performing two-dimensional orthogonal transform coding on image data of each block, and compressing the image data, the image data being stored in a frame memory. A face area extracting means for extracting a face area from the image signal; an area setting means for setting a plurality of areas in a stepwise manner in order to perform a filtering process around the extracted face area; The filter coefficients step by step
The image coding apparatus according to claim and Turkey and a filtering means for varying the allowed by the filtering process.