JP3081658B2 - Image signal encoding device and image signal decoding 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 processing apparatus for compressing an image signal and reproducing image data, and more particularly to an image signal encoding apparatus and an image signal decoding apparatus for suppressing generation of noise when encoding and decoding image data. Related to the device.

[0002]

2. Description of the Related Art Generally, when an image signal captured by a solid-state imaging device represented by a CCD (Charge Coupled Device) or the like is stored as digital data in a storage device such as a memory card, a magnetic disk, or a magnetic tape, the data is stored. The amount will be enormous. In order to store such data in a limited storage capacity range, it is necessary to perform some highly efficient compression.

As a highly efficient image data compression method, a method utilizing orthogonal transform coding as disclosed in Japanese Patent Application Laid-Open No. 62-196990 is generally widely known. With reference to FIG. 6, this orthogonal transform coding method will be described.

First, image data (f) is obtained from a solid-state imaging device or the like.
Is input (1), the image data (f) is divided into blocks of a predetermined size to obtain a value (fb) (2).
For each of the divided blocks, a two-dimensional D
CT (Discrete Cosine Transform) is performed to convert to a value (F) (3). Next, linear quantization according to each frequency component is performed (4), and this quantized value (FQ) is subjected to Huffman coding as variable length coding (5), and the result is compressed data (C ) Is transmitted or recorded. At this time, the quantization width of the linear quantization is prepared by preparing a quantization matrix representing a relative quantization characteristic in consideration of visual characteristics for each frequency, and multiplying the quantization matrix by a constant to reduce the quantization width. I have decided.

On the other hand, when reproducing image data from compressed data, a quantized value (FQ) of a transform coefficient is obtained by decoding (decoding) a variable length code (C). It is impossible to obtain the previous true value (F), and the result obtained by the inverse quantization is a value including an error (F
') (7). Therefore, IDCT for this value
(8), the resulting value (fb ′) is inversely blocked (9), and the resulting image data (f ′) also includes an error. Therefore, the reproduced image (f ′) reproduced and output by the image reproducing device or the like has deteriorated image quality. That is, the error in the value (F ′) obtained as a result of the inverse quantization causes deterioration in the image quality of the reproduced image (f ′) as a so-called quantization error.

The above operation will be specifically described with reference to FIG. First, as shown in FIG. 7A, one frame of image data is divided into blocks of a predetermined size (for example, blocks A, B, C,... Composed of 8 × 8 pixels), and the divided Two-dimensional DCT is performed for each block as orthogonal transform, and the blocks are sequentially stored on an 8 × 8 matrix.

When image data is captured on a two-dimensional plane,
It has a spatial frequency which is frequency information based on the distribution of density information. Therefore, by performing the DCT, the image data is converted into a DC component DC and an AC component AC as shown in FIG. 7B, and the origin position ((0, 0) is displayed on the 8 × 8 matrix. Data indicating the value of the DC component DC is stored at ()), and data indicating the maximum frequency value of the AC component AC in the horizontal axis direction is stored at (0, 7). The (7,0) position has an AC component AC in the vertical axis direction.
The data indicating the maximum frequency value of the AC component AC in the oblique direction is stored at the (7, 7) position. Further, at the intermediate position, the frequency data in the direction related by each coordinate position is stored in such a manner that the frequency data having the higher frequency than the origin side appears.

Next, linear quantization corresponding to each frequency component is performed by dividing data stored at each coordinate position in this matrix by a quantization width for each frequency component. Huffman coding is performed as variable length coding. At this time, regarding the DC component DC, the difference value between the DC component and the DC component of the neighboring block is Huffman-coded.
With respect to the AC component AC, scanning is performed from a low frequency component to a high frequency component called a zigzag scan, and the continuous number of invalid (value is “0”) components (zero run number) and the succeeding valid component Two-dimensional Huffman encoding of the values of the components is performed to obtain encoded data.

In this system, the compression ratio is generally controlled by changing the quantization width of the quantization. The higher the compression ratio, the larger the quantization width.
Therefore, the quantization error becomes large, and the deterioration of the image quality of the reproduced image becomes noticeable.

[0010] The quantization error of the transform coefficient tends to appear mainly as two types of distortion in a reproduced image.
One of them is so-called block distortion in which discontinuity occurs at a block boundary portion, in which a boundary of a block that did not exist in the original image becomes visible. The other is
This is called mosquito noise that occurs because the effect of the quantization error of the DCT coefficient for each block appears in the entire block by the inverse transform. This mosquito noise is D
Because of the quantization error of each sequence of CT coefficients,
It may contain a very low frequency component to a high frequency component, and generally, a hazy pattern tends to appear around a strong luminescent spot or near an edge.
These two types of distortion are visually noticeable, so even if S
Even if / N is good, the subjective impression becomes worse.

Then, the image reproduced by the decoder is
As a distortion removal process, a method of applying a low-pass (low-pass) filter has been considered. This post-filter can remove distortion including high frequencies relatively well.

[0012]

However, the above-described distortion removal method is effective for a system including relatively high frequency components such as distortion due to discontinuity of a block boundary, but is effective for mosquito noise. However, there is a disadvantage that distortion that changes slowly in the block cannot be removed.

[0013] The visibility of the viewer is higher than the luminance signal.
Since the sensitivity of the color difference signal is low, in general image compression, the amount of information allocated to the color difference signal is made smaller than that of the luminance signal to increase the compression ratio. Therefore, generally, the quantization step of the color difference signal is larger than that of the luminance signal, and the generated distortion amount is also large. In particular, mosquito noise tends to appear as low-frequency noise, and consequently has a problem in that it becomes noticeable as a color blur or a colored pattern in an image.

Accordingly, the present invention suppresses the generation of mosquito noise of a color difference signal that is perceived as color blur or colored pattern noise in an image reproduced by encoding and decoding a highly compressed image signal. It is an object to provide an image signal encoding device and an image signal decoding device.

[0015]

In order to achieve the above object, the present invention provides an image signal separating means for separating an input image signal into a luminance signal and a color difference signal, and a luminance separated and output by the image signal separating means. A pixel determination unit that determines whether a pixel of interest is a saturated pixel from the signal value, and a steep change in the color difference signal value of the pixel of interest determined as a saturated pixel by the pixel determination unit with the color difference signal values of surrounding pixels To relax
A chrominance signal distortion suppressing unit for suppressing distortion, a chrominance signal subjected to distortion suppression processing by the chrominance signal distortion suppressing unit,
It is possible to provide an image signal encoding device including a transform encoding unit that encodes and outputs the luminance signal.

A decoding means for decoding the coded input image data into a luminance signal and a chrominance signal; and determining whether a pixel of interest is a saturated pixel based on the value of the luminance signal decoded by the decoding means. Pixel determining means for determining, signal value changing means for changing the color difference signal value of the pixel of interest determined to be a saturated pixel by the pixel determining means to an appropriate value, a color difference signal output from the signal value changing means, An image signal decoding apparatus comprising: a synthesizing unit that synthesizes a luminance signal and outputs the image signal as an image signal.

[0017]

According to the image signal encoding apparatus and the image signal decoding apparatus having the above-described configurations,

At the time of encoding an image signal, the color difference signal of a pixel determined to be a "saturated pixel" is replaced with the value of a color difference signal "0" of the "saturated pixel" by the value of a peripheral color difference signal which is not a "saturated pixel". A distortion suppression process is performed to mitigate a steep change in the color difference signal.

At the time of decoding, the value of the color difference signal of a pixel determined to be a "saturated pixel" is changed to "0". After that, the changed color difference signal and the luminance signal are respectively converted into Y / C signals.
The signal is synthesized by the synthesizing circuit 18 and output as a reproduced signal.

[0020]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of an image encoding device according to an embodiment of the present invention.

First, an input image signal is separated by a Y / C separation circuit 10 into a luminance signal Y and color difference signals Cr and Cb. The luminance signal Y is sent to the transform coding circuit 13 and the luminance signal pixel determination circuit 11.

The luminance signal pixel determination circuit 11 compares the luminance value of each pixel with two threshold values,
Detects "overexposure" and the other th2 detects "underexposure", which is determined based on the following conditions. Y ≧ th1, Y ≦ th2 (1) where th1> th2.

A pixel satisfying the conditional expression is determined as a "saturated pixel" which is "overexposed" or "underexposed". However, in the present embodiment, Y is 8-bit data representing a luminance signal, and is set to th1 = 230 and th2 = 15. Next, a signal is sent to the color difference signal distortion suppression processing circuit 12 in order to perform a distortion suppression process on the color difference signal of the pixel determined as the “saturated pixel”.

In the color difference signal distortion suppression processing circuit 12,
A sudden change in the color difference signal is mitigated so as to suppress the occurrence of distortion. Specifically, the color difference of the color difference signal at the same position as the “saturated pixel” has a value of almost “0”, but this is replaced with the value of the surrounding color difference signal that is not “saturated pixel”. . The result of performing this distortion suppression process on each of Cr and Cb is sent to the transform coding circuit 13 in the same manner as the luminance signal, and is compressed and output. An example in which such distortion suppression processing is simplified in one dimension will be described with reference to FIGS. 2A to 2D.

In FIG. 2A, a bright spot exists while the luminance signal is slowly changing, and it is determined as "overexposed" by comparison with the threshold value th1. The color difference signal C at the same position has a value close to "0" in the overexposed portion as shown in FIG. When the chrominance signal C is compression-encoded and decoded / reproduced, the steep change becomes dull as shown in FIG. 2C, and a change occurs in the originally flat portion. The shaded portion in this figure appears as a color blur or a colored pattern. This tends to be more remarkable as the change is sharper and the contrast is larger.

Therefore, the color difference signal of the overexposed portion is replaced with the color difference signal of the non-overexposed portion in the vicinity thereof, as shown in FIG.
(D). Such a signal hardly changes even if it is encoded and decoded with a considerably high compression ratio. That is, portions other than the overexposed portions are faithfully reproduced. Next, a block diagram of the configuration of the image signal decoding device is shown and described in the block diagram of FIG.

First, in the chrominance signal encoded and decoded into the signal as shown in FIG. 2D, almost no distortion occurs except for the overexposed portion, but the overexposed portion has the actual value. Is very different. However, it has been found that the true value of this part was almost "0".
Therefore, in order to obtain a “saturated pixel” of “overexposure” or “underexposure” for the luminance signal decoded by the decoding circuit 15, first, the luminance signal is converted to the luminance signal pixel determination circuit 1.
6 and a “saturated pixel”
Is determined. The decoding circuit 1 guided to the color difference signal changing circuit 17 based on the determination result
The value of the pixel determined as the “saturated pixel” of the color difference signal output from 5 is changed to “0”.

Thereafter, the Y / C synthesizing circuit 18 synthesizes the chrominance signal whose value has been changed by the chrominance signal changing circuit 17 and the luminance signal output from the decoding circuit 15 and outputs the reproduced signal. I do.

The distortion suppressing means used in the image signal encoding apparatus according to the present invention may use a color difference signal of a pixel determined to be a saturated pixel as a pixel of interest (for example, 7 ×) in addition to the above. 7) replacing the average value of the color difference signals of the non-saturated pixels in the window, or replacing the value of the target pixel with the value obtained by extrapolation or interpolation from the color difference signals of the neighboring non-saturated pixels. No problem. It is preferable to use a low-pass filter together with the above-described distortion suppression processing of the present embodiment.

As described above, since the color fringing and the colored pattern become remarkable due to the sharp change of the color difference signal, the degree of the change is relaxed.
Distortion can be suppressed.

As an example, as shown in FIG. 4, a low-pass filter 23 is connected to the output of the color difference signal distortion suppression processing circuit 22 so as to perform low-pass filtering. In this example, a Y / C separation circuit 20, a luminance signal pixel determination circuit 21, and a color difference signal distortion suppression processing circuit 2
2 and the transform coding circuit 24 have the same function as in the present embodiment. At this time, the luminance signal pixel determination circuit 21
Sends the information of the saturated pixel to the low-pass filter 23, and according to the information, the low-pass filter 23
Although the configuration is such that the pixels to be subjected to the filtering process can be determined, it is needless to say that filtering may be performed uniformly over the entire screen. The effect of this filtering is shown in FIG.
This will be described with reference to FIG.

FIGS. 5A and 5B are examples in which the luminance signal and the color difference signal of an image in which a black tree branch extends on a blue sky background are represented in one dimension. By comparing the luminance signal Y with the threshold value th2, the branch portion is determined to be "blackout". According to the determination result, the result C ′ obtained by performing the distortion suppression processing as described in the embodiment on the color difference signal C is shown in FIG.
This is shown in FIG.

This is because, regarding the contrast of the color difference,
Since pixels which have been reduced but have not been determined as “blackout” at the very end of the threshold due to the threshold processing remain, a ripple-like change point may appear. Therefore,
Distortion still occurs. However, FIG.
As shown in (1), this ripple-like change point can be almost removed by a low-pass filter.

That is, by performing the processing of both the color difference signal distortion suppression processing circuit 22 and the low-pass filter 23, the high contrast and steep change points in the color difference signal can be almost removed. The low-pass filter used at this time may be of such a size as to reduce the ripple-like change as shown in FIG. 5 (c). it can.

The present invention is not limited to the above-described embodiment. The definition of the luminance signal and the chrominance signal is defined by Y, I, and Q as used in a television signal, or expressed by chromaticity coordinates. Y, u, and v, or intensity, hue, and saturation that simulate human visual information may be used.

The data compression method is optional, and can be applied to vector quantization, the DPCM method, and a moving image compression system in addition to transform coding. Of course, various modifications and applications are possible without departing from the spirit of the invention.

[0037]

As described above in detail, according to the present invention, a mosquito of a color difference signal which is perceived as color blur or colored pattern noise in an image reproduced by encoding an image signal and decoding image data. An image signal encoding device and an image signal decoding device capable of suppressing generation of noise can be provided.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating an example in which distortion suppression processing is simplified in one dimension.

FIG. 3 is a block diagram illustrating a configuration of an image decoding device according to an embodiment of the present invention.

FIG. 4 is a block diagram illustrating a configuration of an image encoding device to which a low-pass filter is added as an embodiment of the present invention.

FIG. 5 is an example in which a luminance signal and a color difference signal of an image in which a black tree branch extends on a background of a blue sky are one-dimensionally represented.

FIG. 6 is a block diagram showing an example of a conventional orthogonal transcoding of a compression method for image data.

FIG. 7 is a diagram illustrating an example in which one frame of image data is divided into blocks of a predetermined size.

[Explanation of symbols]

10: Y / C separation circuit, 11: luminance signal pixel determination circuit,
12: color difference signal distortion suppression processing circuit, 13: conversion encoding circuit, Y: luminance signal, Cr, Cb: color difference signal, 15: decoding circuit, 16: luminance signal pixel determination circuit, 17: color difference signal changing circuit, 18 ... Y / C synthesis circuit, 20 ... Y / C separation circuit, 21 ... Luminance signal pixel determination circuit, 22 ... Color difference signal distortion suppression processing circuit, 23 ... Low pass filter, 24 ... Transform coding circuit.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 11/04 G06T 9/00 H04N 1/41 H04N 7/30

Claims (2)

(57) [Claims] An image signal separating unit that separates an input image signal into a luminance signal and a color difference signal; and a pixel that determines whether a pixel of interest is a saturated pixel based on a value of the luminance signal separated by the image signal separating unit. Judgment means, a color difference signal distortion suppression means for reducing the steep change of the color difference signal value of the pixel of interest determined as a saturated pixel by the pixel judgment means with the color difference signal values of surrounding pixels, and suppressing distortion, An image signal encoding apparatus, comprising: a transform encoding unit that encodes and outputs a color difference signal subjected to distortion suppression processing by the color difference signal distortion suppression unit and the luminance signal. 2. A decoding means for decoding coded input image data into a luminance signal and a chrominance signal, and determining whether a pixel of interest is a saturated pixel based on a value of the luminance signal decoded by the decoding means. Pixel determining means for determining, a signal value changing means for changing the color difference signal value of the pixel of interest determined to be a saturated pixel by the pixel determining means to an appropriate value, a color difference signal output from the signal value changing means, An image signal decoding device, comprising: synthesizing means for synthesizing the luminance signal and outputting as an image signal.