DE102006028734A1 - Reduction method for block artifacts from multiple images, involves interpolating image pixels, which results image block as function of determined texture direction of image pixels - Google Patents

Abstract

The method involves interpolating image pixels, which results an image block (2) as a function of a determined texture direction of the image pixels, built up from image blocks. Each image block contains a matrix of image pixels with color values and brightness values. A cross fade factor (k) is determined from the similar color values selected by brightness values of image pixels. The selected color values of picture pixels are interpolated as a function of the cross fade factor. Independent claims are also included for the following: (1) a circuit arrangement for the reduction of block artifacts from multiple images built up from image blocks (2) a digital camera with an image sensor.

Description

The The invention relates to a method for reducing block artifacts from among a plurality of images composed of image blocks, wherein each image block is a matrix of image pixels with color values and brightness values contains and an interpolation of image pixels each of an image block in dependence from a determined texture direction of the image pixels.

• – einem Zeilenspeicher für jeweils eine Zeile von Farb- und/oder Helligkeitswerten von Bildpixeln eines Bildblocks,
• – einen aus dem Zeilenspeicher kontinuierlich mit Tripeln von drei in einer Reihe oder Spalte nebeneinander liegenden Bildpixeln beschreibbarem Matrixspeicher,
• – eine Textur-Detektionseinheit am Ausgang des Matrixspeichers zur Ermittlung eines Texturindikators in horizontaler und vertikaler Richtung aus den im Matrixspeicher zwischengespeicherten Bildpixeln, und
• – einen Interpolator am Ausgang des Matrixspeichers, um in Abhängigkeit eines ermittelten Texturindikators die Bildpixel eines im Matrixspeicher zwischengespeicherten Bildblocks zu interpolieren.

The invention further relates to a circuit arrangement for reducing blocking artifacts of images composed of a plurality of image blocks using such a method, the circuit arrangement comprising:

• A line memory for one row of color and / or brightness values of image pixels of an image block,
• A matrix memory writable from the line memory continuously with triplets of three image pixels lying next to one another in a row or column,
• A texture detection unit at the output of the matrix memory for determining a texture indicator in the horizontal and vertical direction from the image pixels buffered in the matrix memory, and
• An interpolator at the output of the matrix memory in order to interpolate the image pixels of an image block buffered in the matrix memory as a function of a determined texture indicator.

The The invention further relates to a digital camera with an image sensor with a 5 × 3 Sensor matrix for capturing image blocks with a corresponding Bayer mosaic arranged color information, the image sensor is coupled to the circuit arrangement described above.

In digital image processing, the interpolation of color and brightness values of image pixels from image blocks is a well-known and solved problem in itself. In particular, charge-coupled image sensors (charge coupled device CCD) used in image cameras require such interpolation. Circuits and methods for interpolation are for example in the EP 1 289 310 A2 , U.S. Patent 4,630,307 and US 2003/0095703 A1.

Out U.S. Patent 6,091,862 discloses a circuit arrangement and method for pixel interpolation of arranged in a so-called Bayer mosaic image blocks described. The interpolation result is determined by weighting factors for individual Pixel improves, depending on be determined by the texture direction of the image pixels.

Out U.S. Patent 6,549,233 B1 is a color interpolation technique for images known in the Bayer mosaic, in the case of an image pixel of an image block, the minimum one and maximum gradient of the image pixel values. For this The minimum and maximum intensity values of the image pixels become one Image block determined and averaged.

From the EP 0 632 633 A2 It is known to determine the texture direction (gradients) from the difference of luminance values in the vertical and horizontal image directions and to compare them with a threshold value. The interpolation of additional luminance values then tracks in the direction of the detected gradient.

As well For example, US Pat. No. 5,373,322 describes an interpolation method. determined at the gradient (texture direction of image pixels) and the Interpolation depending on performed by the texture direction become. The interpolation is linear.

the The interpolation method described is basically the The problem is that with digital cameras image data via sensor arrays be recorded, which at each picture position only one of the possible three color components (red, green and blue). The missing color components of the color dots have to subsequently over the so-called de-mosaic process can be reconstructed to a complete Color image with red and green and blue information. The sensors in a digital camera sensor and the so existing color information in the output image of the digital camera sensor are often arranged in the so-called Bayer mosaic pattern, which a certain arrangement of red, green and blue information in a 5x3 matrix describes.

At the De-mosaic process will be the missing color information of the image interpolated, being conspicuous and disturbing Errors (artifacts) can occur. These depend on the algorithm used and the image content. Especially disturbing to the viewer are so-called feather or fringe artifacts that interpolate occur at edges.

outgoing It is an object of the present invention to provide an improved To provide a method of reducing block artifacts, in which the interpolation of selected ones Color values is improved.

• – Bestimmen von mindestens einem Überblendungsfaktor aus der Ähnlichkeit von Helligkeitswerten von Bildpixeln, und
• – Interpolation ausgewählter Farbwerte von Bildpixeln in Abhängigkeit von dem Überblendungsfaktor.

The object is achieved by the method of the type mentioned above by:

• Determining at least one blending factor from the similarity of brightness values of image pixels, and
• Interpolation of selected color values of image pixels as a function of the blending factor.

Instead of the linear interpolation used so far is proposed with one of the similarity blending factor formed by the brightnesses of pixels between the chrominance values of those used for the interpolation Apply pixels. This ensures that for pixels same luminance similar Chrominance values result. In the interpolation of the color values Thus, the luminance is taken into account.

hereby In addition, interpolation errors are added based on the assumption that reduces the chrominance values locally considered to be associated with certain luminance values and similar luminance values in similar Chrominance values result.

color artifacts can to be even better reduced.

The crossfade factor is preferably the difference between the absolute difference value the brightness value of the image pixel in the center of the image block matrix and a first further brightness value of an adjacent image pixel and the absolute difference value of the brightness value of the image pixel in the center of the image block matrix and a second further brightness value of an adjacent image pixel.

So For example, for a horizontal texture, the absolute difference value with the horizontal to the image pixel in the center of the image block matrix adjacent values and for vertical texture the absolute difference value with the vertical to the image pixel in the center of the image block matrix adjacent values are calculated.

Also the determined difference may preferably be a brightness base value, preferably the value 128, are added up to the crossfade factor to obtain.

The Texture direction can, for example, in horizontal and / or vertical Direction of the image blocks be determined. This will be diagonal or other texture directions not considered.

The For example, texture direction may be a direction of uniform structures be in the image content, which is determined in a conventional manner.

• a) Interpolation von Bildpixeln eines Bildblocks in horizontaler Richtung bei einer ermittelten horizontalen Textur;
• b) Interpolation von Bildpixeln eines Bildblocks in vertikaler Richtung bei einer ermittelten vertikalen Textur; und
• c) Interpolation von Bildpixeln eines Bildblocks in horizontaler und vertikaler Richtung, wenn keine Textur oder eine horizontale und vertikale Textur erkannt wurde.

It is advantageous if the interpolation takes place in a manner known per se in a direction dependent on the direction of the texture. The following cases can be distinguished:

• a) interpolation of image pixels of an image block in the horizontal direction with a determined horizontal texture;
• b) interpolation of image pixels of an image block in the vertical direction with a determined vertical texture; and
• c) interpolation of image pixels of an image block in the horizontal and vertical directions when no texture or horizontal and vertical texture has been detected.

The Determination of the texture direction can be done, for example, by calculation the sum of absolute differences between color or brightness values take place of image pixels of an image block. That is, that color or Compare brightness values of image pixels of an image block and from the differences the texture is determined.

For example may be a determination of an indicator for the texture direction depending from the color of the image pixel in the center of the image block matrix respectively. The middle image pixel thus gives the center of the to be determined Texture direction and forms the base point for the determination of the texture distinctive color or brightness differences of the image pixels.

A horizontal texture direction can be calculated, for example an associated indicator as the sum of the absolute differences each in the horizontal direction next image pixel values for the luminance or the color green are calculated when the image pixel is at the center of the image block is a chrominance, a red, or a blue component.

The vertical texture direction can be detected with an indicator as the sum of the absolute differences of each in vertical Direction next Image pixel values for the luminance or the color green is calculated when the image pixel is at the center of the image block is a chrominance, red or blue component.

A certain texture direction is recognized, for example, if the corresponding indicator for the horizontal or vertical texture direction a given Threshold exceeds.

The Method is preferred for Picture blocks in the so-called Bayer mosaic with an image block size of 5 × 3 image blocks used.

The object is further achieved by a circuit arrangement of the type mentioned above in that the texture detection unit further for determining at least one cross-fading factor from the similarity of brightness values of image pixels and the interpolator is designed to interpolate selected color values of image pixels as a function of the blending factor.

Of the Interpolator is, for example, by programming or appropriate hardware is set up to be the one described above Performs method.

The Invention will now be described with reference to an embodiment with the accompanying drawings explained in more detail. It demonstrate:

1 Block diagram of a circuit for reducing block artifacts;

2 - 5 × 3 Bayer image block with red, blue and green color values of image pixels and blue component in the center;

3 - 5 × 3 Bayer image block with red, blue and green color values of image pixels and green component in the center.

The 1 shows a block diagram of a circuit arrangement 1 to reduce block artifacts from images made up of a plurality of image blocks 2 are composed. In a line memory 3 For example, the color or brightness values of the image pixels are collected by a camera sensor, such as the red, green, or blue components of a Bayer mosaic, each 1 × 8 bits. The line memory 3 saves the image pixels for two lines. At the output of the line memory 3 are vertical triples (3 × 8 bits) consisting of the existing color values of three superimposed pixels of an image block 2 to a matrix memory 4 delivered. In each clock cycle, such a triple is passed. From a 5x3 matrix stored in matrix memory 4 is stored (5 × 3 × 8 bits) with a texture detection unit 5 examined if the image block 2 has a texture in the horizontal and / or vertical direction. For this purpose, texture indicators are generated, which are connected to a delay element 6 delayed by one clock cycle (Z ^-1 ) to an interpolator 7 be transmitted. The matrix memory 4 provides a complete 5 × 3 matrix for interpolation in each clock cycle. To calculate the texture indicators of a cycle in advance and to improve the timing is the output of the matrix memory 4 is designed as a 6x3 matrix, ie three lines with six pixels. The interpolation uses the right 5x3 values of the output matrix and the texture detection uses the left 5x3 matrix values. The left matrix corresponds to the right matrix for the next pixel.

The texture detection unit 6 is furthermore for determining at least one cross-fading factor k from the similarity of brightness values of image pixels of an image block 2 educated. The crossfade factor k is also sent to the interpolator 7 forwarded.

The interpolator receives the 5 × 3 matrix (5 × 3 × 8 bits) one frame at a time 2 from the matrix memory 4 and performs interpolation in response to the texture indicator T and the blending factor to complete image data for a pixel in RGB format (3 × 8 bits) at the output of the circuit 1 to deliver.

The Interpolation is explained below.

For each pixel are first in a 5 × 3 matrix of an image block 2 So-called textures, ie in the horizontal or vertical direction uniform structures in the image content recognized. The texture information is included in the form of a texture indicator T in the subsequent interpolation, which in turn is performed on a 5x3 matrix of an image block. As a result, artifacts are prevented or at least reduced in particular on edges and lines in the image content. Despite the complexity of the calculation, only two line memories are required for the entire interpolation process.

The 5x3 matrix of an image block 2 each represents a section of an input image. The color value in the middle of a 5 × 3 matrix of an image block 2 belongs to the pixel currently being interpolated. From the surrounding points, the color information is arranged according to the Bayer mosaic. There are two different ones in the 2 and 3 illustrated cases for the calculation depending on the existing color of the current pixel in the middle of the image block 2 ,

The 2 leaves a 5 × 3 matrix of an image block with the color value B ₂₃ in the middle of the image block 2 detect. Furthermore, according to the Bayer mosaic, green (G) and red (R) color values are present.

Comparable with the illustrated case A is also the case not sketched, if there is red information in the middle of the image block is.

The Indian 3 Case B shown leaves an image block 2 recognize where in the middle of the image block 2 green information G _{23 is present} in a vertical row of red information R ₁₃ , R ₃₃ .

The Calculation of a green component with existing blue and red component, as well as the interpolation of red and blue with green component present in red or blue blue row is identical in each case and therefore also in each case considered as a case.

With the texture detection unit is examined in a first step, whether horizontal and vertical textures are present. For this purpose is the presence of only small detail changes in horizontal or vertical direction detected. The information about existing texture directions is for used the interpolation, which is either horizontal, vertical or bilinear. Bilinear interpolation is only possible in carried out the case, if no texture was found or a texture in both horizontal and vertical direction.

By the interpolation in a certain direction depending on of the detected texture indicator T, the known spring or fringe artifacts when interpolating on edges prevented or at least reduced.

Missing green information in case A becomes exclusive with the inclusion of the green information of the dots in the 5x3 matrix of an image block 2 calculated by the color point B ₂₃ . The calculated green component is then used to estimate the brightness (luminance), since in the RGB color space the color green has the largest share of the brightness. Blue and red values are not calculated directly on the basis of the surrounding information of the corresponding color, but on the basis of the difference of this color to the luminance value, the chrominance.

In case B ( 3 ) is the green component G ₂₃ and thus the luminance already exists. The missing red and blue components R ₂₃ and B _{23 are} calculated directly. This again takes place via the chrominance information.

in the Standard mode known per se becomes missing red and blue color values calculated by linear interpolation of chrominances, e.g. For example, the differences from red and blue to the green value, which is used here as luminance.

in the extended according to the invention Mode, interpolation errors are additionally based on the assumption reduces the chrominance values locally with certain ones Luminance values are related and similar luminance values in similar Chrominance values result.

Input data of the method are single 8-bit values of red, green or blue color for image pixels, depending on the sample position in the Bayer pattern. The circuit arrangement 1 Outputs an interpolated frame of 24-bit RGB pixels with the same resolution as the input image.

For detecting existing textures in the texture detection unit 5 First, a horizontal and a vertical texture indicator T are determined, the evaluation then provides information about a present texture. The texture indicators T are calculated as the sum of absolute differences of certain color values in the respective direction. Mostly green information is used. Since the G brightness can be estimated via green G.

The calculation of the texture indicators T is dependent on the color of the pixel in the center of an image block 2 , In the case of an existing R or B component (case A) and the case of a green component in the center of the image block (case B), the indicators are calculated as follows:
Case A: T H = abs (G 11 - G 13 ) + abs (G 13 - G 15 ) + abs (G 22 - G 24 ) + abs (G 31 - G 33 ) + abs (G 33 - G 35 ) T v = abs (G 11 - G 31 ) + abs (G 13 - G 33 ) + abs (G 15 - G 35 ) Case B: T H = abs (G 12 - G 14 ) + abs (G21 - G23) + abs (G 23 - G 2 ) + abs (G 32 - G 34 ) Green in blue row: T v = abs (G 12 - G 23 ) + abs (R 13 - R 33 ) + abs (G 11 - G 34 ) Green in red row: T v = abs (G12 - G32) + abs (B 13 - B 33 ) + abs (G 14 - G 24 ).

Therein, T _{h is} a texture indicator for a horizontal texture T _{v is} a texture indicator for the vertical texture. A vertical texture. A horizontal or vertical texture is detected in each case by comparing the texture indicators T _h and T _v with a given threshold and against each other:
Horizontal texture: (T H + Threshold <T v ) Vertical texture: (T v + Threshold <T H )

The interpolation takes place depending on a recognized horizontal or vertical or no recognized texture. In the following, the embodiments are limited to the description for the case A with existing blue information and for the case B with existing green information in a picture line with existing blue color values. In the case of the existing red information and the green information, the interpolation on identical but with replacement of red and green components performed.

The missing green values be considered a recognized texture direction linear interpolated. Otherwise, the green color is interpolated bilinearly.

The Chrominance is determined by linear or bilinear interpolation of the difference the red and green value calculated.

• a) horizontale Textur: G₂₃= (G₂₂ + G₂₄)/2 R₁₃ = (R₁₂ + R₁₄)/2 R₃₃ = (R₃₂ + R₃₄)2 R₂₃ = G₂₃ + (R₁₃ – G₁₃)/2 + (R₃₃ – G₃₃)/2 = G₂₃ + (R₁₂ + R₁₄ + R₃₂ + R₃₄)/4 – (G₁₃ + G₃₃)/2
• b) vertikale Textur G₂₃ = (G₁₃ + G₃₃)/2 R₂₂ = (R₁₂ + R₃₂)/2 R₂₄ = (R₁₄ + R₃₄)/2 R₂₃ = G₂₃ + (R₂₂ – G₂₂)/2 + (R₂₄ – G₂₄)/2 = G₂₃ + (R₁₂ + R₁₄ + R₃₂ + R₃₄)/4 – (G₂₂ + G₂₄)/2
• c) keine Textur erkannt: G₂₃ = (G₁₃ + G₃₃ + G₂₂ + G₂₄)/4 R₂₃ = (R₁₂ + R₁₄ + R₃₂ + R₃₄)/4.

Case A (described in the case of existing blue information):

• a) horizontal texture: G ₂₃ = (G ₂₂ + G ₂₄ ) / 2 R ₁₃ = (R ₁₂ + R ₁₄ ) / 2 R ₃₃ = (R ₃₂ + R ₃₄ ) 2 R ₂₃ = G ₂₃ + (R ₁₃ - G ₁₃ ) / 2 + (R ₃₃ - G ₃₃ ) / 2 = G ₂₃ + (R ₁₂ + R ₁₄ + R ₃₂ + R ₃₄ ) / 4 - (G ₁₃ + G ₃₃ ) / 2
• b) vertical texture G ₂₃ = (G ₁₃ + G ₃₃ ) / 2 R ₂₂ = (R ₁₂ + R ₃₂ ) / 2 R ₂₄ = (R ₁₄ + R ₃₄ ) / 2 R ₂₃ = G ₂₃ + (R ₂₂ - G ₂₂ ) / 2 + (R ₂₄ - G ₂₄ ) / 2 = G ₂₃ + (R ₁₂ + R ₁₄ + R ₃₂ + R ₃₄ ) / 4 - (G ₂₂ + G ₂₄ ) / 2
• c) no texture detected: G ₂₃ = (G ₁₃ + G ₃₃ + G ₂₂ + G ₂₄ ) / 4 R ₂₃ = (R ₁₂ + R ₁₄ + R ₃₂ + R ₃₄ ) / 4.

• a) horizontale Textur: G₁₃ = (G₁₂ + G₁₄)/2 G₃₃ = (G₃₂ + G₃₄)/2 R₂₃ = G₂₃ + (R₁₃ – G₁₃)/2 + (R₃₃ – G₃₃)/2 = G₂₃ + (R₁₃ + R₃₃)/2 – (G₁₂ + G₁₄ + G₃₂ + G₃₄)/4 B₂₃ = (B₂₂ + B₂₄)/2
• b) vertikale Textur: G₂₂ = (G₁₂ + G₃₂)/2 G₂₄ = (G₁₄ + G₃₄)/2 R₂₃ = (R₁₃ + R₃₃)/2 B₂₃ = (B₂₂ – G₂₂)/2 + (B₂₄ – G₂₄)/2 = G₂₃ + (B₂₂ + B₂₄)/2 – (G₁₂ + G₁₄ + G₃₂ + G₃₄)/4
• c) keine Textur erkannt: R₂₃ = G₂₃ + (R₁₃ + R₃₃)/2 – (G₁₂ + G₁₄ + G₃₂ + G₃₄)/4 B₂₃ = G₂₃ + (B₂₂ + B₂₄)/2 – (G₁₂ + G₁₄ + G₃₂ + G₃₄)/4

Case B (described in the case of a green color value in blue image line):

• a) horizontal texture: G ₁₃ = (G ₁₂ + G ₁₄ ) / 2 G ₃₃ = (G ₃₂ + G ₃₄ ) / 2 R ₂₃ = G ₂₃ + (R ₁₃ - G ₁₃ ) / 2 + (R ₃₃ - G ₃₃ ) / 2 = G ₂₃ + (R ₁₃ + R ₃₃ ) / 2 - (G ₁₂ + G ₁₄ + G ₃₂ + G ₃₄ ) / 4 B ₂₃ = (B ₂₂ + B ₂₄ ) / 2
• b) vertical texture: G ₂₂ = (G ₁₂ + G ₃₂ ) / 2 G ₂₄ = (G ₁₄ + G ₃₄ ) / 2 R ₂₃ = (R ₁₃ + R ₃₃ ) / 2 B ₂₃ = (B ₂₂ - G ₂₂ ) / 2 + (B ₂₄ - G ₂₄ ) / 2 = G ₂₃ + (B ₂₂ + B ₂₄ ) / 2 - (G ₁₂ + G ₁₄ + G ₃₂ + G ₃₄ ) / 4
• c) no texture detected: R ₂₃ = G ₂₃ + (R ₁₃ + R ₃₃ ) / 2 - (G ₁₂ + G ₁₄ + G ₃₂ + G ₃₄ ) / 4 B ₂₃ = G ₂₃ + (B ₂₂ + B ₂₄ ) / 2 - (G ₁₂ + G ₁₄ + G ₃₂ + G ₃₄ ) / 4

The Interpolation takes place in an extended mode according to the invention for selected color values of image pixels in dependence from a crossfade factor k. This mode is for case A) with blue information available and in case with existing green information described in an image line with existing blue color values. The missing green values are identical to the standard mode described above linear or interpolated bilinearly. In the case of no recognized texture, the chrominance calculation is also identical to the standard mode.

in the Trap of a texture is used in a chrominance calculation cross-fading the linear interpolation between the chrominance values of those used for the interpolation Pixels applied. The crossfade factor k becomes from the similarity the brightnesses of the pixels formed, so for pixels same luminance similar Chrominance values result.

• a) horizontale Textur: G₂₃ = (G₂₂ + G₂₄)/2 R₁₃ = (R₁₂ + R₁₄)/2 R₃₃ = (R₃₂ – R₃₄)/2 k = abs(G₁₃ – G₂₃) – abs(G₃₃ – G₂₃) – 128 R₂₃ = G₂₃ + (1 – k)·(R₁₃ – G₁₃) + k·(R₃₃ – G₃₃) = G₂₃ + R₁₃ – G₁₃ + k·(R₃₃ – G₃₃ – R₁₃ + G₁₃)
• b) vertikale Textur: G₂₃ = (G₁₃ + G₃₃)/2 R₂₂ = (R₁₂ + R₃₂)/2 R₂₄ = (R₁₄ + R₃₄)/2 k = abs(G₂₂ – G₂₃) – abs(G₂₄ – G₂₃) + 128 R₂₃ = G₂₃ + (1 – k)·(R₂₂ – G₂₂) + k·(R₂₄ – G₂₄) = G₂₃ + R₂₂ – G₂₂ + k·(R₂₄ – G₂₄ – R₂₂ + G₂₂)
• c) keine Textur erkannt: G₂₃ = (G₁₃ + G₃₃ + G₂₂ + G₂₄)/4 R₂₃ = (R₁₂ + R₁₄ + R₃₂ + R₃₄)/4

Case A (described in the case of existing blue information):

• a) horizontal texture: G ₂₃ = (G ₂₂ + G ₂₄ ) / 2 R ₁₃ = (R ₁₂ + R ₁₄ ) / 2 R ₃₃ = (R ₃₂ - R ₃₄ ) / 2 k = abs (G ₁₃ - G ₂₃ ) - abs (G ₃₃ - G ₂₃ ) - 128 R ₂₃ = G ₂₃ + (1 - k) · (R ₁₃ - G ₁₃ ) + k · (R ₃₃ - G ₃₃ ) = G ₂₃ + R ₁₃ - G ₁₃ + k · (R ₃₃ - G ₃₃ - R ₁₃ + G ₁₃ )
• b) vertical texture: G ₂₃ = (G ₁₃ + G ₃₃ ) / 2 R ₂₂ = (R ₁₂ + R ₃₂ ) / 2 R ₂₄ = (R ₁₄ + R ₃₄ ) / 2 k = abs (G ₂₂ - G ₂₃ ) - abs (G ₂₄ - G ₂₃ ) + 128 R ₂₃ = G ₂₃ + (1 - k) · (R ₂₂ - G ₂₂ ) + k · (R ₂₄ - G ₂₄ ) = G ₂₃ + R ₂₂ - G ₂₂ + k · (R ₂₄ - G ₂₄ - R ₂₂ + G ₂₂ )
• c) no texture detected: G ₂₃ = (G ₁₃ + G ₃₃ + G ₂₂ + G ₂₄ ) / 4 R ₂₃ = (R ₁₂ + R ₁₄ + R ₃₂ + R ₃₄ ) / 4

• a) horizontale Textur: G₁₃ = (G₁₂ + G₁₄)/2 G₃₃ = (G₃₂ + G₃₄)/2 k = abs(G₁₃ – G₂₃) – abs(G₃₃ – G₂₃) + 128 R₂₃ = G₂₃ + (1 – k)·(R₁₃ – G₁₃) + k·(R₃₃ – G₃₃) = G₂₃ + R₁₃ – G₁₃ + k·(R₃₃ – G₃₃ – R₁₃ + G₁₃) B₂₃ = (B₂₂ + B₂₄)/2
• b) vertikale Textur: G₂₂ = (G₁₂ + G₃₂)/2 G₂₄ = (G₁₄ + G₃₄)/2 R₂₃ = (B₁₃ + R₃₃)/2 k = abs(G₂₂ – G₂₃) – abs(G₂₄ – G₂₃) + 128 B₂₃ = G₂₃ +(1 – k)·(B₂₂ – G₂₂) + k·(B₂₄ – G₂₄) = G₂₃ + B₂₂ – G₂₂ + k·(B₂₄ – G₂₄ – B₂₂ + G₂₂)
• c) keine Textur erkannt: R₂₃ = G₂₃ + (R₁₃ + R₃₃)/2 – (G₁₂ + G₁₄ + G₃₂ + G₃₄)/4 B₂₃ = G₂₃ + (B₂₂ + R₂₄)/2 – (G₁₂ + G1₄ + G₃₂ + G₃₄)/4.

Case B (described in the case of a green color value in blue image line):

• a) horizontal texture: G ₁₃ = (G ₁₂ + G ₁₄ ) / 2 G ₃₃ = (G ₃₂ + G ₃₄ ) / 2 k = abs (G ₁₃ - G ₂₃ ) - abs (G ₃₃ - G ₂₃ ) + 128 R ₂₃ = G ₂₃ + (1 - k) · (R ₁₃ - G ₁₃ ) + k · (R ₃₃ - G ₃₃ ) = G ₂₃ + R ₁₃ - G ₁₃ + k · (R ₃₃ - G ₃₃ - R ₁₃ + G ₁₃ ) B ₂₃ = (B ₂₂ + B ₂₄ ) / 2
• b) vertical texture: G ₂₂ = (G ₁₂ + G ₃₂ ) / 2 G ₂₄ = (G ₁₄ + G ₃₄ ) / 2 R ₂₃ = (B ₁₃ + R ₃₃ ) / 2 k = abs (G ₂₂ - G ₂₃ ) - abs (G ₂₄ - G ₂₃ ) + 128 B ₂₃ = G ₂₃ + (1 - k) · (B ₂₂ - G ₂₂ ) + k · (B ₂₄ - G ₂₄ ) = G ₂₃ + B ₂₂ - G ₂₂ + k · (B ₂₄ - G ₂₄ - B ₂₂ + G ₂₂ )
• c) no texture detected: R ₂₃ = G ₂₃ + (R ₁₃ + R ₃₃ ) / 2 - (G ₁₂ + G ₁₄ + G ₃₂ + G ₃₄ ) / 4 B ₂₃ = G ₂₃ + (B ₂₂ + R ₂₄ ) / 2 - (G ₁₂ + G 1 ₄ + G ₃₂ + G ₃₄ ) / 4.

Claims (17)

Method for reducing block artifacts of a plurality of image blocks ( 2 ) composite images, each image block ( 2 ) contains a matrix of image pixels with color values and brightness values and an interpolation of image pixels in each case of an image block ( 2 ) takes place as a function of a determined direction of the texture of the image pixels, characterized by Determining at least one blending factor (k) from the similarity of brightness values of image pixels, and interpolating selected color values of image pixels as a function of the blending factor (k). A method according to claim 1, characterized in that the texture direction in the horizontal and / or vertical direction of the image blocks ( 2 ) is determined. Method according to claim 1 or 2, characterized that the texture direction is a direction of uniform structures in the picture content is. Method according to one of the preceding claims, characterized characterized in that the interpolation is in one of the texture direction dependent direction he follows. Method according to one of the preceding claims, characterized by - interpolation of image pixels of an image block ( 2 ) in the horizontal direction with a detected horizontal texture; Interpolation of image pixels of an image block ( 2 ) in the vertical direction with a detected vertical texture; and interpolation of image pixels of an image block ( 2 ) in horizontal and vertical directions if no texture or horizontal and vertical texture was detected. Method according to one of the preceding claims, characterized by determining the texture direction by calculating the sum of absolute differences between color or brightness values of image pixels of an image block ( 2 ). Method according to one of the preceding claims, characterized by determining a texture direction indicator in dependence from the color of the image pixel in the center of the image block matrix. Method according to Claim 7, characterized by calculating an indicator for the horizontal texture direction as the sum of the absolute differences of the respective horizontally adjacent image pixel values for the luminance or the color green when the image pixel center point of the image block ( 2 ) is a chrominance, red or blue component. A method according to claim 7 or 8, characterized by calculating an indicator for the vertical texture direction as the sum of the absolute differences of the respective vertically next image pixel values for the luminance or the color green, if the image pixel in the center of the image block ( 2 ) is a chrominance, red or blue component. Method according to one of Claims 7 to 9, characterized by calculating an indicator for the vertical texture direction as the sum of the absolute differences of the respectively vertically adjacent image pixel values for the luminance or color green and for the vertical column in the middle of the image block ( 2 ) as the sum of the absolute differences of the next vertical image pixel values for the chrominance or the colors red or green. Method according to one of the preceding claims, characterized characterized in that a texture direction is detected when the determined texture direction each exceeds a defined threshold. Method according to one of the preceding claims, characterized by calculating the blending factor from the difference between the absolute difference value of the brightness value of the Image pixels in the center of the image block matrix and a first further Brightness value of an adjacent image pixel as well as the absolute one Difference value of the brightness value of the image pixel in the center of Image block matrix and a second further brightness value of a adjacent image pixels. Method according to claim 12, characterized in that that on the difference a brightness base value, preferably the Value 128, is added to the blending factor (k) to obtain. A method according to claim 12 or 13, characterized by calculating the absolute difference values with the horizontal adjacent to the image pixel in the center of the image block matrix Horizontal texture values and calculating the absolute difference values with the vertical to the image pixel in the center of the image block matrix adjacent vertical texture values. Circuit arrangement ( 1 ) for reducing block artifacts from a plurality of image blocks ( 2 ) composite images with the method according to one of the preceding claims, comprising: - a line memory ( 3 ) for one line of color and / or brightness values of image pixels of an image block ( 2 ), - one from the line memory ( 3 ) continuously with triples of three in a row or column adjacent image pixels recordable matrix memory ( 4 ), - a texture detection unit ( 5 ) at the output of the matrix memory ( 4 ) for determining a texture indicator (T) in the horizontal and vertical direction from those in the matrix memory ( 4 ) cached Image pixels, and - an interpolator ( 7 ) at the output of the matrix memory ( 4 ), in dependence on a determined texture indicator (T), the image pixels of one in the matrix memory ( 4 ) cached image block ( 2 ) to interpolate. characterized in that - the texture detection unit ( 5 ) for determining at least one cross-fading factor (k) from the similarity of brightness values of image pixels and - the interpolator ( 7 ) is designed for the interpolation of selected color values of image pixels as a function of the blending factor (k). Circuit arrangement ( 1 ) according to claim 15, characterized by a delay element ( 6 ) at the output of the texture detection unit, to the interpolator ( 7 ) forwarded texture indicator (T) to delay at least one clock cycle. Digital camera with an image sensor with a 5 × 3 sensor matrix for capturing image blocks ( 2 ) arranged according to a Bayer mosaic color information, characterized in that the image sensor with the circuit arrangement ( 1 ) is connected according to claim 15 or 16.