Methods for object-oriented image sharpness recovery

Abstract

Method for recovering the sharpness of an original image (1) in which the original image (1) is converted into a result image (5) such that a. from the intensity relief (2) of the original image (1) a long-drawn-out ripple component (3) and a short-drawn-out ripple component (4) are obtained, which b. are independently converted into an intensity base range (8) and two intensity boundary ranges (9, 10) of the intensity scale (7) of the result image (5), wherein c. the elongated ripple component (3) of the intensity relief (2) of the original image (1) is assigned to a region in the middle of the intensity scale (7) of the result image (5), which represents its intensity base region (8), and d. the shortened ripple component (4) of the intensity relief (2) of the original image (1) is converted into the intensity boundary regions (9, 10) of the intensity scale (7) of the result image (5) such that e. the negative values ​​of the short-wave ripple component (4) the lower intensity boundary region (9) before the intensity base region (8) in the intensity scale (7) of the result image (5) and f. the positive values ​​of the short-wave ripple component (4) are assigned to the upper intensity boundary region (9) according to the intensity base region (8) in the intensity scale (7) of the result image (5), characterized in that a separation of the intensity relief (2) of the original image (1) into a long-wave ripple component (3) and a short-wave ripple component (4) is carried out using the Christo method with a virtual scanning element (11), wherein all intensity values ​​of an original image (1) are converted into the intensity values ​​of a result image (5), such that the converted intensity values ​​are calculated linearly, proportionally to the ratio of the sizes of corresponding intensity ranges of the intensity relief (2) of the original image (1) and the intensity scale (6) of the result image (5), as follows: I d i f f _ m i n u s _ j = I b a s i c _ b o t − I s h o r t _ m i n u s _ j ∗ I d i f f _ m i n u s _ t o p − I d i f f _ m i n u s _ b o t I s h o r t _ m i n u s _ t o p − I s h o r t _ m i n u s _ b o t , I b a s i c _ j = I b a s i c _ b o t + I l o n g _ j ∗ I b a s i c _ t o p − I b a s i c _ b o t I l o n g _ t o p − I l o n g _ b o t , I d i f f _ p l u s _ j = I b a s i c _ t o p + I s h o r t _ p l u s _ j ∗ I d i f f _ p l u s _ t o p − I d i f f _ p l u s _ b o t I s h o r t _ p l u s _ t o p − I s h o r t _ p l u s _ b o t , wobei: I diff_minus_j ein Intensitätswert im unteren Intensitätsrandbereich (9) vor dem Intensitätsbasisbereich (8) der Intensitätsskala (7) des Ergebnisbildes (5), I short_minus_j ein negativer Wert aus dem kurzgezogenen Welligkeitsanteil (4) des Intensitätsreliefs (2) des Originalbildes (1), I short_minus_botthe lower limit of the negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (5), I short_minus_top the upper limit of the negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I basic_j an intensity value in the base intensity range of the intensity scale (7) of the result image (5), I long_j a value from the elongated ripple component (3) of the intensity relief (2) of the original image (1), I long_bot the lower limit of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I long_top the upper limit of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I diff_plus_j an intensity value in the upper intensity range (10) according to the intensity base range (8) of the intensity scale (7) of the result image (5), I short_plus_j a positive value from the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I short_plus_bot the lower limit of the positive values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I short_plus_top the upper limit of the positive values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1).

Description

[0001] The invention relates to a method for restoring the sharpness of an image, which is adjusted according to the minimum degree of detail to be sharpened in this image.

[0002] A blurry image can occur for various reasons. For example, the objects being photographed may have been located outside the focal plane or at the edge of the image. This can happen in both long-range (geo-sensing) and close-up (microscopy) imaging. Images obtained with non-conventional imaging techniques, such as infrared, terahertz, ultrasound, X-ray, or similar systems, also exhibit significant blurring. These types of applications have become increasingly important in recent decades. The blurring of an infrared (IR) image is caused by the excessively long wavelength of the IR radiation. In contrast, blurring of X-ray images results from the extremely high penetrating power of this radiation, leading to a weak contrast between the photographed objects and their background.Furthermore, the aforementioned images appear in a significantly wider dynamic range than human perception can perceive. Therefore, they must be converted to an 8-bit format. This process results in a loss of sharpness in these images, and consequently, the loss of many captured objects in the converted image. For this reason, image sharpness recovery is of considerable economic importance in many scientific and industrial fields.

[0003] Various state-of-the-art methods exist that influence an image by modifying its histogram (contrast stretching) to improve its sharpness (Contrast and Sharpness / Christoph Heckenkampf. - Inspekt, No. 4, pp. 34-36, 2013). However, these methods alter the contrast of the entire image without considering pixel proximity (so-called homogeneous pixel operation). Consequently, contrast stretching cannot guarantee a true increase in sharpness, which, moreover, only works as a neighborhood operation (inhomogeneous pixel operation) (Digital Image Processing / Bernd Jähne. - 4th ed. - Berlin, Heidelberg, etc.: Springer, 1997).

[0004] Furthermore, various edge filters in diverse forms for image contrast enhancement are described in the literature (Digital Image Processing / Bernd Jähne. - 4th ed. - Berlin, Heidelberg, etc.: Springer, 1997). All these methods offer some image improvement, but either fail to enhance small details of the image under investigation or also amplify the noise in the entire image. In these cases, the image under investigation is often binarized, meaning that no fine image information is reproduced in the "enhanced" image.

[0005] Next, the high- and low-frequency components of an image under investigation could be calculated separately and then combined into a single image, thus amplifying the high-frequency portion, e.g., using an "unsharp mask" filter (Photoshop). The same approach is also proposed for the dynamic range conversion of a high dynamic range image (F. Branchitta, M. Diani, G. Corsini, M. Romagnolli - New technique for the visualisation of high dynamic range infrared images. Opt. Eng., Vol. 48, 096401, 2009). However, filtering the raw data, e.g., with a high-pass or low-pass filter or with the Sawitzky-Golay filter, which tolerates rapid changes in the signal and still smooths it, by no means eliminates all the noise present in the signal (Interpreting the Stars Correctly / Andreas Hofer and Lothar Wenzel. - Elektronik 10, pp. 92-101, 2001).The use of filters such as the Gaussian filter or the so-called exponential filter also reaches its limits (Modelling Telecommunication Components: Limits of the Gaussian Approximation - PHOTONICS SPECTRA, pp. 127-133, May 2002) and leads to misinterpretations, as a signal can be incorrectly reproduced with overshoot. Therefore, a clean separation between the long-wave ripple component and the short-wave ripple component of the signal is neither stable nor reliable with conventional filters. Furthermore, this method is prone to generating various artifacts.

[0006] It is also known that several disadvantages of conventional filters can be partially avoided by using so-called morphological filtering (Morphological Image Processing / P. Soille. - Berlin, Heidelberg: Springer, 1998). Morphological filtering refers to the processing of various points of the object under investigation depending on its immediate surroundings. Morphological filters represent effective tools (transformations) that, especially in combination, are suitable for solving various problems. For example, a combination of erosion and dilation can be used to smooth the resulting curve to varying degrees (K. Subr, C. Soler, F. Durand - Edge-preserving multiscale image decomposition based on local extrema. ACM Transactions on Graphics (TOG), 2009, Vol. 28, No. 5, p. 147). This envelope filter performs proper neighborhood operations and directly affects the sharpness of an image.However, the application of such a filter is quite limited. For example, it will not offer any way to influence the specific size of the image detail of interest (Chapter 2.2). Furthermore, this method also proves to be quite time-consuming and complicated.

[0007] A simple combination of the corresponding envelope filters, such as erosions and dilatations (J. Kumar, MS Shunmugam - A new approach for filtering of surface using morphological operations - International Journal of Machine Tools and Manufacture, 2006, Vol. 46, No. 3, pp. 260-270), which are known from edge and surface analysis, cannot, in general, guarantee a reliable separation between the long-wave and short-wave components of a signal. Furthermore, the intensity outliers that arise randomly using this method can be mishandled. This will inevitably lead to the production of artifacts.

[0008] The invention is based on the objective of creating a method that enables the simple, precise, and stable recovery of image sharpness. This method should also be applicable as an automatic and flexible process, which is set up without any changes to the source code, but rather through simple parameterization on a user interface.

[0009] The solution to the technical problem is provided by the features of claim 1.

[0010] According to the invention, image sharpness is restored using an object-oriented method in which the sharpness of the captured image is enhanced depending on the size of the details of interest. This effect can be compared to the mechanical focusing of a microscope. For this purpose, a long-wave ripple component and a short-wave ripple component are extracted from the intensity relief of the original image. These components are independently converted into different intensity ranges on the intensity scale of a resulting image. The long-wave ripple component of the original image's intensity relief is assigned to a region in the center of the resulting image's intensity scale. The short-wave ripple component of the original image's intensity relief is converted into two intensity ranges at the extreme ends of the resulting image's intensity scale.The negative values ​​of the short-wave ripple component of the intensity relief in the original image are assigned to the lower intensity boundary region before the base intensity region of the intensity scale in the resulting image. Conversely, the positive values ​​of the short-wave ripple component of the intensity relief in the original image are assigned to the upper intensity boundary region after the base intensity region of the intensity scale in the resulting image.

[0011] According to claim 1, the intensity relief of the original image is separated into a long-wave ripple component and a short-wave ripple component using the Christo method (see Chapter 3, R. Louban, Image Processing of Edge and Surface Defects, Springer, 2009), which is based on the principle of mechanically scanning and enveloping a curve under investigation with a virtual scanning element (structural element). Accordingly, the long-wave ripple component is determined as the center line of the corresponding multiple scan curves. The short-wave ripple component is then calculated as the difference between the original profile and the determined long-wave ripple component.Thus, these components can be cleanly and reliably separated and then processed independently, ensuring that even small intensity deviations, much smaller than the absolute amplitude fluctuations of the original image's intensity relief, are reliably detected and evaluated. All intensity values ​​of the original image are linearly converted into the intensity values ​​of the resulting image, proportionally to the ratio of the sizes of the corresponding intensity ranges in the original image's intensity relief to the intensity scale of the resulting image. Ibasic_j=Ibasic_bot+Ilong_j∗Ibasic_top−Ibasic_botIlong_top−Ilong_bot, Idiff_minus_j=Ibaisc_bot−Ishort_minus_j∗Idiff_minus_top−Idiff_minus_botIshort_minus_top−Ishort_minus_bot, Idiff_plus_j=Ibasic_top+Ishort_plus_j∗Idiff_plus_top−Idiff_plus_botIshort_plus_top−Ishort_plus_bot, where: I diff_minus_jthe individual intensity value in the lower intensity range before the base intensity range of the intensity scale of the result image, I short_minus_j the single negative value from the short-drawn ripple component of the intensity relief of the original image, I short_minus_bot the lower limit of the negative values ​​of the short-wave ripple component of the intensity relief of the original image, I short_minus_top the upper limit of the negative values ​​of the short-wave ripple component of the intensity relief of the original image, I basic_j the individual intensity value in the base intensity range of the intensity scale of the result image, I long_j the individual value from the elongated ripple component of the intensity relief of the original image, I long_bot the lower limit of the elongated ripple component of the intensity relief of the original image, I long_topthe upper limit of the elongated ripple component of the intensity relief of the original image, I diff_plus_j the individual intensity value in the upper intensity range according to the intensity base range of the intensity scale of the result image, I short_plus_j the single positive value from the short-wave ripple component of the intensity relief of the original image, I short_plus_bot the lower limit of the positive values ​​of the short-wave ripple component of the intensity relief of the original image, I short_plus_top the upper limit of the positive values ​​of the short-wave ripple component of the intensity relief of the original image.

[0012] Unlike conventional filtering methods, this procedure does not distort the characteristic components of the curve under investigation, thus allowing for error-free curve analysis.

[0013] Furthermore, the limit values ​​of the intensity ranges of the long- and short-wave ripple components of the intensity relief of the original image can be determined using various standard image processing methods, e.g., the CatEye2 method (High-Dynamic-Renge (HDR) Vision / Berndt Hoefflinger (Ed.) - Springer series in Advanced Microelectronics, Vol. 26, Springer, Berlin, Heidelberg, New York, 2007, pp. 112-114). Thus, an original image can be quickly, easily, and reliably converted into a result image with restored sharpness.

[0014] According to claim 2, the size of the virtual scanning element used to perform the Christo process is defined as larger than the minimum dimension of the details to be sharpened, but smaller than the elongated ripples. Thus, the size of the virtual scanning element determines how finely the short-wave ripple component (curve deviation component) of the intensity relief of the original image is calculated and, consequently, how detailed the sharpening of the original image content is in the resulting image.

[0015] The virtual scanning element (structural element) of the Christo process can be designed in various shapes depending on requirements. For example, an original image whose intensity values ​​are integers can be processed very quickly with a virtual scanning element designed as a cuboid with a corresponding side width. Conversely, an original image with intensity values ​​as floating-point numbers can be processed very finely and precisely with a virtual scanning element in the form of a sphere with a corresponding diameter.

[0016] According to claim 3, the limit values ​​of the intensity base range and the intensity boundary ranges of the intensity scale of a result image are defined based on the maximum value of its intensity scale and the sharpness enhancement factor as follows: Idiff_minus_bot=0, Idiff_minus_top=Ibasis_bot=I0∗0.5∗(1−ξ), Ibasic_top=Idiff_plus_bot=I0∗0.5∗(1+ξ), Idiff_plus_top=I0, where: I0 is the maximum value of the intensity scale of the result image, which has I0 grayscale levels (from 0 to I0), ξ the sharpness enhancement factor, I diff_minus_bot the lower limit of the lower intensity range of the intensity scale of the result image for negative values ​​of the short-wave ripple component of the intensity relief of the original image, I diff_minus_top the upper limit of the lower intensity boundary range of the intensity scale of the result image for negative values ​​of the short-wave ripple component of the intensity relief of the original image, I basis_bot the lower limit of the intensity base range of the intensity scale of the result image for the intensity values ​​of the elongated ripple component of the intensity relief of the original image, I basic_topthe upper limit of the intensity base range of the intensity scale of the result image for the intensity values ​​of the elongated ripple component of the intensity relief of the original image, I diff_plus_bot the lower limit of the upper intensity range of the intensity scale of the result image for positive deviations of the short-wave ripple component of the intensity relief of the original image, I diff_plus_top the upper limit of the upper intensity boundary range of the intensity scale of the result image for positive values ​​of the short-wave ripple component of the intensity relief of the original image.

[0017] The individual values ​​of the long-wave ripple component of the intensity relief in the original image are positive or zero, while the individual intensity values ​​of the short-wave ripple component can be positive, negative, or zero. In this way, the determined ripple components of the original image are independently converted into the corresponding ranges of the intensity scale of the resulting image. Thus, the sharpness information of the processed original image is reliably and stably extracted and enhanced in the resulting image.

[0018] According to claim 4, the intensity factor ξ, whose value lies in the range [0.0 - 1.0], is determined empirically and subsequently used as a typical, task-specific value. Thus, the sharpness of an original image can be easily and variably enhanced in practice.

[0019] An intensity factor ξ value of 0.6 can serve as a fairly universally suitable setting, ensuring a stable and balanced sharpness enhancement of an original image under a wide variety of shooting conditions.

[0020] To adapt the intensity scale of the result image to human perception, the intensity scale of a result image can conveniently be set as an 8-bit scale, which has 256 gray levels (from 0 to 255), where I0 = 255.

[0021] According to claim 5, the individual intensity values ​​from the short-wavelength component of the intensity relief of the original image are non-linearly converted into the intensity values ​​of the corresponding intensity boundary regions of a resulting image, depending on their sign. The larger the values ​​of the short-wavelength component of the intensity relief of the original image, the closer these values ​​are positioned to the outer edge of their respective intensity boundary regions. This allows the sharpness of an original image to be enhanced even more significantly. Such a non-linear conversion can be implemented using various standard mathematical methods, such as polynomial power, exponential, or similar functions.

[0022] The details of the invention, as well as its further features, applications and advantages, are described in the following exemplary embodiments with reference to the Fig. Figures 1-6 are explained. All described or illustrated features, whether individually or in any combination, constitute the subject matter of the invention, irrespective of their compilation in the claims or their cross-references, and irrespective of their formulation or representation in the description or in the drawings. The following are shown: Fig. 1 Schematic representation of a circuit board, recorded with an electronic microscope, as original image (a) and as result image (b), generated using the registered method for object-oriented recovery of the sharpness of an image. Fig. 2 Schematic representation of a parking lot, recorded with an infrared (IR) camera, as original image (a) and as result image (b), generated using the registered method for object-oriented image sharpness recovery, including image dynamic conversion from a 16-bit scale to an 8-bit scale. Fig. 3. Schematic representation of the intensity relief of an original image and its ripple components, as well as the intensity relief of a resulting image. To simplify the representation, all intensity vectors that are actually two-dimensional are depicted as one-dimensional vectors. Fig. 4 Schematic representation of a conversion of the values ​​of the determined elongated ripple component of the intensity relief of an original image into the intensity base range of the intensity scale of a result image. Fig. 5 Schematic representation of a conversion of the values ​​of the determined short-wave ripple component of the intensity relief of an original image into the intensity boundary areas of the intensity scale of a result image. Fig. 6 Schematic representation of the division of the intensity scale of a result image into a base intensity range and two edge intensity ranges, which were calculated with a sharpness enhancement factor ξ=0.6.

[0023] As an example, a photograph of a circuit board taken with an electronic microscope ( Fig. 1, a) and an image of a parking lot taken with an IR camera are used. The intensity values ​​of the first image are in an 8-bit range, which corresponds to human perception, while the second image contains intensity values ​​in the 16-bit range. To make an infrared original image visible, it is converted into an 8-bit image using the CatEye2 method (High-Dynamic-Renge (HDR) Vision / Berndt Hoefflinger (Ed.) - Springer series in Advanced Microelectronics, Vol. 26, Springer, Berlin, Heidelberg, New York, 2007, P. 112-114) ( Fig. 2, a). In both recordings, rather blurry images were produced, probably for different reasons, which are processed in the same way according to the Christo method described above with a virtual scanning element (11).

[0024] From an intensity relief (2) of an original image (1), an elongated ripple component (3) is determined using the method described above with a virtual scanning element (11) with a dimension of 2 pixels, as is the case in Fig. 3, a and b is shown schematically. A short-wave ripple component (4) of the original image (1) is calculated as a difference between the real intensity vector (2) and its long-wave ripple component (3) ( Fig. 3, b).

[0025] The elongated ripple component (3) of the intensity relief (2) of the original image (1) is assigned to a region in the middle of the intensity scale (7) of a result image (5), which represents an intensity base region (8) of the intensity scale (7) of the result image (5) ( Fig. 4).

[0026] The negative values ​​of the short-wave ripple component (4) of the intensity relief (2) of the original image (1) are assigned to the lower intensity boundary region (9) before the intensity base region (8) of the intensity scale (7) of the result image (5). Conversely, the positive values ​​of the short-wave ripple component (4) of the intensity relief (2) of the original image (1) are assigned to the upper intensity boundary region (10) after the intensity base region (7) of the intensity scale (7) of the result image (5). Fig. 5) An intensity relief (6) of the resulting image (5) clearly shows both the intensity profile and the characteristic intensity differences, and thus all sharpness differences of the original image (1). The latter can even be variably enhanced.

[0027] The conversion of the intensity values ​​of the original image (1) into the intensity values ​​of the result image (5), which has an 8-bit intensity scale, was performed linearly with the intensity factor ξ=0.6 ( Fig. 6).

[0028] This results in a fundamental difference between the method used and conventional filters. The latter use minimizing the cumulative deviation of an approximate curve from the curve under investigation (Gaussian method) as a criterion for calculating a long-wave ripple component (3). Under certain circumstances, this can erroneously lead to some overshoot or losses of the signal under investigation, since the scattering of this signal naturally also corresponds to a Gaussian distribution.

[0029] In contrast, the registered method uses an envelope around the curve under investigation. This takes into account a maximum permissible deviation of individual values ​​of the curve under investigation from its determined elongated ripple component (Chebyshev method) (Pocketbook of Mathematics / IN Bronstein, KA Semendjajew, G. Musiol, H. Mühling. - 5th ed. - Thun and Frankfurt am Main: Harri Deutsch Verlag, 2001). Consequently, the registered method provides a result image (5) with a significant recovery of image sharpness without any misinterpretations ( Fig. 1,b and 2,b).

[0030] In summary, the proposed method ensures object-oriented image sharpness recovery and provides a sharp result image that functions simply, precisely, stably, and reliably, regardless of the original image or its scale dimension. Therefore, this method can also be used flawlessly for image dynamic range conversion without any loss of information. Furthermore, the registered method can be applied to automatically sharpen captured images by increasing the sharpness of the captured image according to the size of the details of interest.

Claims

[1] Method for recovering the sharpness of an original image (1) in which the original image (1) is transformed into a result image (5) such that a. from the intensity relief (2) of the original image (1) a long-drawn-out ripple component (3) and a short-drawn-out ripple component (4) are obtained, which b. are independently converted into an intensity base range (8) and two intensity boundary ranges (9, 10) of the intensity scale (7) of the result image (5), wherein c. the elongated ripple component (3) of the intensity relief (2) of the original image (1) is assigned to a region in the middle of the intensity scale (7) of the result image (5), which represents its intensity base region (8), and d. the shortened ripple component (4) of the intensity relief (2) of the original image (1) is converted into the intensity boundary regions (9, 10) of the intensity scale (7) of the result image (5) such that e. the negative values ​​of the short-wave ripple component (4) the lower intensity boundary region (9) in front of the intensity base region (8) in the intensity scale (7) of the result image (5) and f. the positive values ​​of the short-wave ripple component (4) are assigned to the upper intensity boundary region (9) according to the intensity base region (8) in the intensity scale (7) of the result image (5), characterized by, that a separation of the intensity relief (2) of the original image (1) into an elongated ripple component (3) and a shortened ripple component (4) is carried out using the Christo method with a virtual scanning element (11), wherein all intensity values ​​of an original image (1) are converted into the intensity values ​​of a result image (5), such that the converted intensity values ​​are calculated linearly, proportionally to the ratio of the sizes of corresponding intensity ranges of the intensity relief (2) of the original image (1) and the intensity scale (6) of the result image (5), as follows: Idiff_minus_j=Ibasic_bot−Ishort_minus_j∗Idiff_minus_top−Idiff_minus_botIshort_minus_top−Ishort_minus_bot, Ibasic_j=Ibasic_bot+Ilong_j∗Ibasic_top−Ibasic_botIlong_top−Ilong_bot, Idiff_plus_j=Ibasic_top+Ishort_plus_j∗Idiff_plus_top−Idiff_plus_botIshort_plus_top−Ishort_plus_bot, where: I diff_minus_jan intensity value in the lower intensity range (9) before the intensity base range (8) of the intensity scale (7) of the result image (5), I short_minus_j a negative value from the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I short_minus_bot the lower limit of the negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (5), I short_minus_top the upper limit of the negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I basic_j an intensity value in the base intensity range of the intensity scale (7) of the result image (5), I long_j a value from the elongated ripple component (3) of the intensity relief (2) of the original image (1), I long_bot the lower limit of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I long_top the upper limit of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I diff_plus_j an intensity value in the upper intensity range (10) according to the intensity base range (8) of the intensity scale (7) of the result image (5), I short_plus_j a positive value from the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I short_plus_bot the lower limit of the positive values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I short_plus_top the upper limit of the positive values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1). [2] Method according to claim 1, characterized by , that a measure of the virtual scanning body (11) of the Christo process is set larger than the minimum dimension of the details to be sharpened of an original image (1). [3] Method according to claim 1, characterized by , that the limit values ​​of the intensity base range (8) and the intensity boundary ranges (9, 10) in the intensity scale (7) of a result image (5) are defined as follows: Idiff_minus_bot=0, Idiff_minus_top=Ibasis_bot=I0∗0.5∗(1−ξ), Ibasic_top=Idiff_plus_bot=I0∗0.5∗(1+ξ), Idiff_plus_top=I0, where: I0 is the maximum value of the intensity scale (7) of the result image (5), which has I0 grayscale levels (from 0 to I0), ξ the sharpness enhancement factor, I diff_minus_bot the lower limit of the lower intensity boundary range (9) of the intensity scale (7) of the result image (5) for negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I diff_minus_topthe upper limit of the lower intensity boundary range (9) of the intensity scale (7) of the result image (5) for negative values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1), I basis_bot the lower limit of the intensity base range (8) of the intensity scale (7) of the result image (5) for the intensity values ​​of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I basic_top the upper limit of the intensity base range (8) of the intensity scale (6) of the result image (5) for the intensity values ​​of the elongated ripple component (3) of the intensity relief (2) of the original image (1), I diff_plus_bot the lower limit of the upper intensity range (9) of the intensity scale (7) of the result image (5) for positive values ​​of the short-wave ripple component (4) of the intensity relief (2) of the original image (1), I diff_plus_topthe upper limit of the upper intensity boundary range (10) of the intensity scale (7) of the result image (5) for positive values ​​of the short-drawn ripple component (4) of the intensity relief (2) of the original image (1). [4] Method according to claims 1 and 3, characterized by , that the intensity factor ξ, whose value lies in the range [0,0 - 1,0], is determined based on experience and is subsequently used as a typical value. [5] Method according to claim 1, characterized by , that the intensity values ​​of a result image (5) in the intensity boundary areas (9, 10) are converted non-linearly, compressed towards the outer edge of the respective intensity area.

Images