DE202010017586U1 - Apparatus for improving digital images - Google Patents

Abstract

Device for improving digital images, which includes an analysis of the image parameters with a subsequent extraction of the brightness component, a structure of a correction filter and an image correction, characterized in that the structure of the correction filter based on a determined projection profile by calculating the amplitudes and directions of the brightness profiles of the image the image areas with a high gradient value are then selected and a series of the local brightness projections are built up in a direction which is determined by the gradient in the selected regions, and the image correction is carried out as follows: the amplitude of the gradient of the image is calculated, the pixels of the image are selected in the local area which lies on the edges, except for the pixels which lie directly on the edges, and the selected areas are corrected using the built-up correction filter rt, where the filter values are set on the basis of the local values of the course amplitude.

Description

The invention relates to a device for improving digital images, which includes an analysis of the image parameters with a subsequent extraction of the brightness component, a structure of a correction filter and an image correction.

The invention relates to the processing of digital images and can be used to reduce or completely eliminate distortions arising from the use of other techniques for improving the quality of photographic and video images, such as sharpening or enhancing image resolution, interference filtration.

• – doppelte Lichtkranze (Überstrahlungen) bei einem zusätzlichen Scharfzeichnen (weiter Halo-Effekt genannt);
• – gleiche Verzerrungen für einige Interpolationsarten (insbesondere bei der häufig gebrauchten bikubischen Interpolation) beim Scharfzeichnen.

At present, various methods for improving the image quality are used in the processing of the digital images. For example, many modern digital photo and video cameras are equipped with digital image processing processors that are used, among other things, to improve image quality. The requirement for mobility as well as the price of the photo and video cameras does not allow to use powerful computing resources. As a result, in many cases limited or simplified processing methods are used, which produce undesirable side effects such as:

• - double light ring (blasts) with an additional sharpening (further called halo effect);
• - Same distortions for some types of interpolation (especially in the commonly used bicubic interpolation) when sharpening.

Well known is a process for improving the quality of an original image with the help of an Adobe Photoshop. In this case, a blur filter is used that allows to reduce distortion that occurs when sharpening the original image.

The disadvantage of this method is a lack of selectivity, because the filter is applied to all image areas. Therefore, with a distortion reduction, a sharpness loss occurs on the entire image.

A method for improving the video image [1] is known in which the distorted image areas are automatically determined and the distortions are corrected. Correcting the distortions is done by calculating matched pixel brightnesses in the distorted image areas.

The disadvantage of this method is that the correction is possible only at the image areas in which the so-called blooming can be seen, d. H. the overflow of charge from the adjacent sensor pixels at higher charge levels.

There is also known a method of correcting a dynamic video image [2]. It provides for the correction of step distortions (so-called posterization or separation of tones) in the originally monotone colored image areas, which, for. B. are caused by the compression of the image.

The disadvantage of this method is that the correction of the double halo is impossible and only stepwise distortions are corrected.

There is still a method of editing original digital images [3] known. This process improves image quality by sharpening the image. The brightness channel is extracted from different color channels and inserted back into the color channels after the brightness increase. To increase the image sharpness, a standard linear filter with a core size of 3 × 3 or 5 × 5 or 7 × 7 pixels is used.

The disadvantage of this method is that due to the application of a filter the distortions in the form of halo effect occur.

The closest prior art to the invention is a method for improving the image quality by sharpening [4], which takes into account the peculiarities of recording and reproducing apparatus. In this process, small high-contrast details such as leaves, hair, grass, etc. are protected from overexposure. Overexposure protection is ensured by blending data from the original image and an image with increased sharpness, thereby altering the valences in the local area around the overexposed objects.

The disadvantage of this method is that only the image areas are corrected with an increased sharpness, which can lead to overexposure or lead. In addition, access to the parameters of the original image must be ensured in order to process the image in the manner mentioned.

It is an object of the invention to provide an apparatus for improving digital images of the type mentioned, which produces an improved quality of the image.

The stated object is solved by the features of claim 1.

This device makes it possible to substantially reduce or completely remove distortions that arise during the processing of the original images. In this case, only one image with the distortions already introduced is used as the starting object. The details of the parameters of the original image and the processing method used are not required.

The quality improvement is achieved by including the analysis of the image parameters, an extraction of the luminance component, a construction of a correction filter and an image correction, and the construction of the correction filter based on the determined projection profile. The amplitudes and the directions of the brightness characteristics of the image are calculated. Subsequently, the image areas are selected with a high level value and a number of the local brightness projections are constructed. The direction of these projections is determined by the gradient in the selected area. The image correction is carried out as follows:
The amplitude of the course of the image is calculated, the pixels of the image in the local area are selected, which lie on the edges, except for the pixels which lie directly at the edges. The selected areas are corrected using the built-in correction filter, with the valencies of the filter adjusted based on the local values of the historical amplitude.

The invention will be explained in more detail with reference to the drawings. Show it:

1 the digital image before editing,

2 the same picture after sharpening (in the picture the halo effect is clearly noticeable),

3 the same image after processing by the new device,
4 . 5 . 6 show local projections of the edges in the direction of the gradient:

4 the local projection of the edge of the source image,

5 the local projection of the edge after sharpening, it is a halo (halo effect) to see and

6 the local projection of the edge after correction of the distortions by the new device.

• I. Analyse des Bildes zur Parametererfassung: 1. Extrahierung der Helligkeitskomponente, 2. Berechnen der Amplituden und der Richtungen der Helligkeitsverläufe auf dem Bild, 3. Auswahl der Bildbereiche mit einem hohen Verlaufswert (Kanten), 4. Aufbau einer Reihe von lokalen Helligkeitsprojektionen in Richtung, die durch den Verlauf in den ausgewählten Bereichen bestimmt ist und 5. Berechnen des ermittelten Projektionsprofils anhand der aufgebauten Reihre der lokalen Projektionen.
• II. Aufbau eines Korrekturfilters anhand des berechneten ermittelten Profils.
• III. Bildkorrektur: 1. Berechnen der Verlaufsamplituden des Bildes, 2. Auswahl der Pixel auf dem Bild, die in unmittelbarer Nähe zu den Kanten liegen, außer den Pixeln, die unmittelbar an den Kanten liegen. Für solche Pixel ist ein Verhältnis ausgeführt: max(abs(grad(M)))/abs(grad(p)) > Thr wobei: p – das Pixel, M – die Punktmenge, die dieses Pixel umgibt und Thr – eine bestimmte Grenze, Konstante bedeuten. 3. Korrektur der gefundenen Bereiche mit dem gebauten Filter, wobei die Wertigkeiten des Filters anhand der lokalen Werte der Verlaufsamplitude eingestellt sind.

• I. Analyse des Bildes zur Parametererfassung: a. Extrahierung der Helligkeitskomponente. Angenommen, dass jedes Pixel des Bildes in Form von drei Farbkomponenten R, G, B existiert. Es ist die Helligkeitskomponente (Y) mit der Formel berechnet: Y = R·0.2989 + G·0.587 + B·0.114 Eine solche Darstellung des Signals und die Berechnung der Helligkeitskomponente ist typisch für die Bearbeitung von Foto- und Videobilder und bekannt. b. Berechnen des Absolutwertes und der Richtungen der Helligkeitsverläufe auf dem Bild: Gx_(a,b) = Y_(a-1,b-1) – Y_(a+1,b-1) + Y_(a-1,b+1) – Y_(a+1,b+1) + 2·Y_(a-1,b) – 2·Y_(a+1,b) Gy_(a,b) = Y_(a-1,b-1) + Y_(a+1,b-1) – Y_(a-1,b+1) – Y_(a+1,b+1) + 2·Y_(a,b-1) – 2·Y_(a,b+1) Ga_(a,b) = Gx_(a,b)·Gx_(a,b) + Gy_(a,b)·Gy_(a,b) Gx – die Komponente des Verlaufs in Richtung x, Gy – die Komponente des Verlaufs in Richtung y, Ga – die Quadrat des Absolutwertes vom Verlauf und a, b – die Koordinaten des Pixels. c. Auswahl der Bildbereiche mit einem hohen Verlaufswert (Kanten), Sortierung der Pixel absteigend nach dem Absolutwert des Verlaufs und die Auswahl der ersten N Pixel: N = 4·(W + H)/(W·H) wobei W, H – die Breite bzw. die Höhe des Bildes in Pixel bedeuten. d. Aufbau einer Reihe von lokalen Helligkeitsprojektionen in Richtung, die durch den Verlauf in den ausgewählten Bereichen bestimmt ist.

The improvement of the output image by the device is as follows:

• I. Analysis of the image for parameter acquisition: 1. Extraction of the brightness component, 2. Calculation of the amplitudes and the directions of the brightness gradients on the image, 3. Selection of the image regions with a high gradient value (edges), 4. Construction of a series of local brightness projections in Direction determined by the course in the selected areas and 5. Calculation of the determined projection profile on the basis of the built-up sequence of the local projections.
• II. Construction of a Correction Filter Based on the Calculated Determined Profile.
• III. Image correction: 1. Compute the gradient of the image, 2. Select the pixels in the image that are in close proximity to the edges, except for the pixels that are directly adjacent to the edges. For such pixels, a ratio is executed: max (abs (grad (M))) / abs (grad (p))> Thr where: p - the pixel, M - the set of points surrounding this pixel and Thr - a definite limit, constant. 3. Correction of the found areas with the built filter, whereby the valences of the filter are adjusted by means of the local values of the progression amplitude.

• I. Analysis of the image for parameter acquisition: a. Extraction of the brightness component. Suppose that each pixel of the image exists in the form of three color components R, G, B. It calculates the brightness component (Y) with the formula: Y = R x 0.2989 + G x 0.587 + B x 0.114 Such a representation of the signal and the calculation of the brightness component is typical for the processing of photo and video images and known. b. Calculation of the absolute value and the directions of the brightness gradients on the image: Gx _{(a, b)} = Y _{(a-1, b-1)} -Y _{(a + 1, b-1)} + Y _{(a-1, b + 1)} -Y _{(a + 1, b + 1)} + 2 * Y _{(a-1, b)} - 2 * Y _{(a + 1, b)} Gy _{(a, b)} = Y _{(a-1, b-1)} + Y _{(a + 1, b-1)} -Y _{(a-1, b + 1)} -Y _{(a + 1, b + 1)} + 2 · Y _{(a, b-1)} - 2 · Y _{(a, b + 1)} Ga _{(a, b)} = Gx _{(a, b)} · Gx _{(a, b)} + Gy _{(a, b)} · Gy _{(a, b)} Gx - the component of the curve in the direction x, Gy - the component of the curve in the direction y, Ga - the square of the absolute value of the curve and a, b - the coordinates of the pixel. c. Selection of image areas with a high gradient value (edges), sorting the pixels in descending order according to the absolute value of the gradient and the selection of the first N pixels: N = 4 * (W + H) / (W * H) where W, H - is the width or height of the image in pixels. d. Construction of a series of local brightness projections in the direction determined by the course in the selected areas.

• e. Berechnen des ermittelten Projektionsprofils anhand der aufgebauten Reihe der lokalen Projektionen. v^avg = Σvⁱ
• II. Aufbau eines Korrekturfilterkerns anhand des berechneten ermittelten Profils. f. Erzeugung eines eindimensionalen Filterprofils P_i = [V^avg _(7+i) + V^avg _(9-i)]/2 i = 2..7 P₁ = 1 g. Zerlegen des eindimensionalen Filters in einem radial-symmetrischen zweidimensionalen Kern.

The determination of such a value from the absolute values Gx _{(a, b)} , Gy _{(a, b)} , which is the maximum for that pixel in the coordinate (a, b) (pixel - from the areas selected in step c Taking into account the sign, there are 4 variants: If then the vector of the brightness projection V collects from the pixels in coordinates abs (Gx)> abs (Gy) and Gx <0 V ⁱ _(1..15) = Y _{(a + 7, b)} .. Y _{(a-7, b)} abs (Gx)> abs (Gy) and Gx> 0 V ⁱ _(1..15) = Y _{(a-7, b)} .. Y _{(a + 7, b)} abs (Gx) <abs (Gy) and Gy <0 V ⁱ _(1..15) = Y _{(a, b + 7)} ..Y _{(a, b-7)} abs (Gx) <abs (Gy) and Gy> 0 V ⁱ _(1..15) = Y _{(a, b-7)} .. Y _{(a, b + 7)}

• e. Calculating the determined projection profile based on the constructed series of local projections. v ^avg = Σv ⁱ
• II. Construction of a Correction Filter Core Based on the Calculated Determined Profile. f. Generation of a one-dimensional filter profile P _i = [V ^avg _{(7 + i)} + V ^avg _(9-i) ] / 2 i = 2..7 P ₁ = 1 g. Disassembling the one-dimensional filter in a radially symmetric two-dimensional core.

• III. Bildkorrektur: h. Berechnen der Verlaufsamplituden des Bildes. Gx_(a,b) = Y_(a-1,b-1) – Y_(a+1,b-1) + Y_(a-1,b+1) – Y_(a+1,b+1) + 2·Y_(a-1,b) – 2·Y_(a+1,b) Gy_(a,b) = Y_(a-1,b-1) + Y_(a+1,b-1) – Y_(a-1,b+1) – Y_(a+1,b+1) + 2·Y_(a,b-1) – 2·Y_(a,b+1) Ga_(a,b) = Gx_(a,b)·Gx_(a,b) + Gy_(a,b)·Gy_(a,b) Gx – die Komponente des Verlaufs in Richtung x, Gy – die Komponente des Verlaufs in Richtung y, Ga – das Quadrat des Absolutwertes vom Verlauf und a, b – die Koordinaten des Pixels. i. Auswahl der Pixel des Bildes im lokalen Bereich, der an den Kanten anliegt, außer den Pixeln, die unmittelbar an den Kanten liegen. Für solche Pixel ist ein Verhältnis ausgeführt: W < 3; W ist mit der Formel berechnet:
  
  wobei: (a, b) – die Koordinaten des Pixels sind. j. Korrektur der gefundenen Bereiche mit einem aufgebauten Korrekturfilterkern, wobei die Wertigkeiten des Filters anhand der lokalen Werte der Verlaufsamplitude eingestellt (parametriert) sind. Bei dieser Variante ist nur das zentrale Element des Filterkerns parametriert, was funktional äquivalent zur Bemischung des originalen Bildes mit einer bestimmten Wertigkeit ist.
  
  wobei: Y – die Helligkeitskomponente des Ausgangsbildes; Y' – die Helligkeitskomponente des korrigierten Bildes; F' – das parametrierte Filter.
  
  α – der Normungsfaktor, der auf solche Weise gewählt ist, dass die Summe der Elemente F' gleich 1 ist.

The matrix F, 15 x 15 elements in size, is filled by calculating the distance of the matrix element from the center and selecting the corresponding values from the one-dimensional profile P with interpolation (eg with the ruler).

• III. Image correction: h. Calculate the gradient of the image. Gx _{(a, b)} = Y _{(a-1, b-1)} -Y _{(a + 1, b-1)} + Y _{(a-1, b + 1)} -Y _{(a + 1, b + 1)} + 2 * Y _{(a-1, b)} - 2 * Y _{(a + 1, b)} Gy _{(a, b)} = Y _{(a-1, b-1)} + Y _{(a + 1, b-1)} -Y _{(a-1, b + 1)} -Y _{(a + 1, b + 1)} + 2 · Y _{(a, b-1)} - 2 · Y _{(a, b + 1)} Ga _{(a, b)} = Gx _{(a, b)} · Gx _{(a, b)} + Gy _{(a, b)} · Gy _{(a, b)} Gx - the component of the curve in the direction x, Gy - the component of the curve in the direction y, Ga - the square of the absolute value of the curve and a, b - the coordinates of the pixel. i. Selection of the pixels of the image in the local area, which is applied to the edges, except for the pixels that lie directly at the edges. For such pixels, a ratio is implemented: W <3; W is calculated using the formula:
  
  where: (a, b) - are the coordinates of the pixel. j. Correction of the found areas with a built-up correction filter kernel, wherein the valencies of the filter are set (parameterized) on the basis of the local values of the progression amplitude. In this variant, only the central element of the filter kernel is parametrized, which is functionally equivalent to the blending of the original image with a certain valence.
  
  where: Y - the brightness component of the output image; Y '- the brightness component of the corrected image; F '- the parameterized filter.
  
  α - the standardization factor chosen in such a way that the sum of the elements F 'is equal to 1.

The device for improving the digital photo and video images allows to significantly reduce the distortion and in some cases completely eliminate it. In this case, only the image with the distortions already introduced is used as the starting object for processing. The details of the parameters of the original image and the processing method used are not required.

State of the art

• [1.] Invention application RU No. 2005 125 325 Bull. No. 5 2007
• [2.] Invention application RU No. 2006 105 328 Bull. No. 28 2007
• [3.] Patent RU No. 2298226 Bull. No. 12 2007r
• [4.] Patent US No. 7268916 (Prototype)

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• RU 2005125325 [0028]
• RU 2006105328 [0028]
• RU 2298226 [0028]
• US 7268916 [0028]

Claims (1)

A digital image enhancement apparatus including analysis of the image parameters with subsequent extraction of the luminance component, a correction filter design, and image correction, characterized in that the structure of the correction filter is determined from a determined projection profile by calculating the amplitudes and directions of the image brightness gradients then, the image areas are selected to have a high gradient value, and a series of the local brightness projections are constructed in a direction determined by the gradient in the selected areas, and the image correction is performed in the following manner: the amplitude of the course of the image is calculated, the pixels of the image are selected in the local area which is adjacent to the edges, except the pixels which are immediately adjacent to the edges, and the selected areas are corrected by means of the constructed correction filter ert, where the weights of the filter are set based on the local values of the progression amplitude.

Images