Method for segmentation of text by colour criterion in process of copying

FIELD: physics, image processing.

SUBSTANCE: invention is related to digital processing of image in process of scanning and copying, and especially to the field of colour and black-and-white text segmentation, when text is automatically extracted from scanned document. According to the present invention, method of text segmentation by colour criterion consists in performance of the following operations: initial image is broken down into non-overlapping units of pixels; new image Z is generated, in which each pixel represents corresponding unit of initial image; in process of scanning serial units are selected from initial image; classification is carried out for current unit by criterion "monochromatic/colour" in space of opposite colours; initial colour unit RGB is broken down into monochromatic units R, G and B; detector of Laplacian-Gaussian edges is applied with specified threshold T to monochromatic unit; number of edges is calculated for each pixel in Z; classification of "text/non-text unit" is performed by comparison of edge number with set threshold C; classified channels are combined, using logical operator OR.

EFFECT: provision of universal approach to preliminary processing of initial document, providing for faster printing of copy and saving of multifunctional device resources.

4 cl, 3 dwg

 

The invention relates to digital image processing in scanning and copying, and particularly to the field of segmentation of color and black and white text, when the text is automatically extracted from the scanned document.

A basic set of procedures, which performs the copy technics, in particular Digital Multifunction Device (MFD) or, in other words, a Multifunction Printer (MFP)includes obtaining a scanned image from the scanner, image processing, and sending them to the built-in printer for printing. IFIs should be able to handle different types of originals. These originals can have different contents, such as text, charts, graphics, pictures; they can be printed on various media such as paper of various qualities; they can be created using other methods, such as grayscale or full-color image. These different types of originals can copying be subject to different influences depending on the capabilities and limitations of specific IFIs, in particular, copies can receive the strip, ambient light, unwanted coloration, offset entities, the change in the spectrum, the appearance of moire, etc. even fixed parameter settings copy Pref is changed to different quality reproduction depending on the type of the original. To resolve this problem and obtain reproductions in accordance with the preferences of the user must be able to adapt the procedures performed by the MFP.

A common approach to solving this problem is to develop algorithms that allow you to control the copying process in the MFP. Many algorithms of this kind are based on segmentation of the page and indexing each segmented area in accordance with the special classification. These algorithms require simultaneous access to the entire image of the page and allow multiple analysis of each pixel. On the other hand, any algorithm applied to the copying process must be able to process the image in blocks, and not to repeat the processing of previously processed blocks. This requirement is impracticable based on the application of the known methods.

On the other hand, it is very important to define the color characteristics of the original, as, for example, the classification (classification) of the image to black-and-white originals would simplify the procedure of processing and printing, in particular, you can only use the cartridge with black paint and save the other colors are not at the expense of quality. The various applications of classification when processing documents described in the patent literature.

Thus, U.S. patent No. 6972866 [1], which is an analog of the claimed invention, describes a method of classifying pixels into one of two categories: - the first of which includes a monochrome pixels, and the other - colored pixels. The color of each of pixels of the same color index in the color space L a'b'. Average color index is determined as a function of the color index pixels in the group.

Published patent application U.S. No. 20020081023 [2], which is the analog prototype of the claimed invention, discloses a method of detecting the color of the text for the image processing in the copy device. In this way provided chunked data processing input image by selecting blocks instead of pixel-by-pixel processing. According to this invention, the objects in the image are identified by selection blocks and are divided into two categories: "text" or "text". Then, for each text object determines the foreground color using the data of the foreground text. After collecting data about the foreground calculate the average foreground color in a color space, for example in the space L a'b'. Using information about the average color, determine whether the text object in black and white or not.

A common shortcoming of the aforementioned analogues having aetsa, each of them solves only part of the challenge in improving the efficiency of processing and printing of documents with mixed content, i.e. text and graphics.

The problem to which the invention is directed, is to develop a universal approach to pre-processing of the source document that provides a quick copy printing and saving the resources of the IFIs.

The main technical result is achieved due to the introduction of an improved method for the segmentation of the text color attribute in the backup process, and the inventive method includes the following operations:

- the original image is divided into non-overlapping blocks of pixels;

create a new image Z, in which each pixel represents the corresponding block of the original image;

- sequentially selected scanning blocks of the original image;

- perform classification for the current block on the basis of "monochrome/color" in the space opposite colors;

- break the original colour block RGB to monochrome blocks of R, G and b;

- apply the detector edges of the Laplacian of Gaussian with a given threshold T to monochrome block;

- calculate the number of edges for each pixel in Z;

- perform classification of the text is new/text block by comparing the number of edges with a given threshold;

- combine the classified channels, using the logical OR operator.

For optimal performance of the proposed method it is important that the classification process on the basis of "monochrome/color" was performed by the following operations:

- convert the image from RGB color space into an image represented in the space opposite colors;

- classify pixels as black or colored on the basis of the distance from the axis intensity in the space opposite colors;

- form the blocks of pixels by dividing the image into non-overlapping squares;

- calculate the color saturation of the pixel block as the sum of distances for all pixels belonging to the block;

- set the threshold of saturation with regard to the learning outcomes of the training sample;

- classify the image as a color, if the color saturation is higher than the specified threshold;

- classify the image as monochrome, if the color saturation does not exceed the specified threshold.

For optimal performance of the proposed method it is advisable that the segmentation of the text on the basis of "monochrome/color" performed by separate R, G, In the channels.

For optimal performance of the proposed method makes sense that the segmentation of the text is ispolnali on random colored background.

For a better understanding of the claimed invention, the following is a detailed description of the sequence of operations involving graphic materials.

Figure 1. The block diagram of the algorithm to perform classification on the basis of "monochrome/color.

Figure 2. The block diagram of the algorithm to perform classification on the basis of "text/no text.

Figure 3. Identifying text on the image of the document.

The inventive method is implemented by performing the following steps:

- determination of the monochrome image is performed in the space opposite colors L a'b' [3], which is defined with three channels - intensity L' and color a' and b':

L'=(R+G+B)/3

the position of the point along the axis L' represents the different intensity levels of gray. Monochrome pixel is defined as

a color pixel is defined as

where a'2+b'2represents the square of the distance from the axis L' to any point (a', b') along the axis L' and determines the distance or threshold from the axis L', above which a pixel is considered as a colour. The function T(L') represents the cylinder. Therefore, all points in the color space L a'b' lying inside the cylinder are considered as monochrome; all points outside Qili the DRA are seen as colored.

Color saturation With(In) of the unit pixels is defined as the sum of the distances for all pixels that belong to the block C. the Image is classified as a color, if(In) is greater than the threshold and the image is classified as monochrome otherwise. The threshold is determined from the training sample.

Figure 1 represents the block diagram of the algorithm to perform classification on the basis of "monochrome/color. 101 selects a block in the scanning of the document. In 102 runs the classification block on the basis of "monochrome/color. If the image is monochrome, 103 block is assigned the label of "monochrome". If the image is monochrome, 104 executes the classification block on the basis of "text/no text. If the image is in color, 105 block is assigned the label "color". 106 executes the RGB split into separate R, G and b images. 107 classification is performed for each of R, G, and the image on the basis of "text/no text. 108 merges classified R, G, and images, using a logical "OR".

Figure 2 is a block diagram of the algorithm to perform classification on the basis of "text/no text. In 201 retrieves the selected block in the scanning of the document. In block 202 is processed by the detector edges of the Laplacian of Gaussian. 203 is vicis is giving the number of edges in the block. 204 checks whether the number of edges specified threshold C. If Yes, 205 block is classified as "text" and set the label text. If not, in block 206 is classified as "no text" and assigned the label "do not text".

Figure 3. represents the results of detection of text on the document image. The detected text areas marked in gray.

The inventive method is industrially applicable for digital image processing in scanning and copying, in particular, for the segmentation of color and black and white text, when the text is automatically extracted from the scanned document, for example, in multi-functional digital devices.

Sources of information

1. U.S. patent No. 6972866.

2. Published patent application U.S. No. 20020081023.

3. .Dabov, A.Foi, V.Katkovnik, and .Egiazarian, "Image denoising by sparse 3D transform-domain collaborative filtering," IEEE Trans. Image Process., vol.16, no.8, pp.2080-2095, August 2007.

1. The method of segmentation of the text color attribute, which consists in performing the following operations:
divide the original image into non-overlapping blocks of pixels;
form a new image of Z in which each pixel represents the corresponding block of the original image;
choose scanning successive blocks of the original image;
performed for the current block classification p is the characteristic of the monochrome/color" in the space opposite colors;
break the original colour block RGB to monochrome blocks of R, G and b;
primeneniya detector edges of the Laplacian of Gaussian with a given threshold T to monochrome block;
calculate the number of edges for each pixel in Z;
perform classification "text/text block by comparing the number of edges with a given threshold;
combine the classified channels, using the logical OR operator.

2. The method according to claim 1, characterized in that the segmentation of the text on the basis of "monochrome/color" performed by separate R, G, In the channels.

3. The method according to claim 1, characterized in that the classification process on the basis of "monochrome/color perform the following operations: convert image from RGB color space into an image represented in the space opposite color;
classify pixels as black or colored on the basis of the distance from the axis intensity in the space opposite color;
form blocks of pixels by dividing the image into non-overlapping squares;
expect the color saturation of the pixel block as the sum of distances for all pixels belonging to the block;
set the threshold of saturation with regard to the learning outcomes of the training sample;
classify the image as a color, if the color saturation exceed the t given threshold;
classify the image as monochrome, if the color saturation does not exceed the specified threshold.

4. The method according to claim 1, characterized in that the segmentation of the text is carried out on a random colored background.



 

Same patents:

FIELD: information technologies.

SUBSTANCE: in color correction scheme per each tint any image which is combination of three different chroma signals is splitted into multiple ranges by tint so that colour correction may be performed by range of each tint. To prevent leaving of uncorrected peripheral range area for colour correction, colour correction procedure is performed so that correction ranges are overlapped with each other.

EFFECT: creation of colour correction procedure performed to provide possibility of true correction of all tints in image without leaving any uncorrected range.

5 cl, 4 dwg

FIELD: office equipment, in particular, system and method for correction of image during output to printing device.

SUBSTANCE: computer, with printing device connected to it, is equipped with information carrier, containing file with image, agent module, data storage module, at least one application, graphic device interface and printing device interface. Agent module is made with possible finding of new file with image on data carrier, aforementioned file containing additional information, calculations for this file of numeric index based on image data and recording of computed index together with additional file information to data storage module. Printing device driver is made with possible calculation of numeric index on basis of additional information data received in it, appropriate to calculated numeric index, and correction of image data received by it and transfer of corrected image data to printing device.

EFFECT: possible correction of image in accordance to additional information during output to printing device from various applications, installed on computer.

2 cl, 1 dwg

FIELD: printing ink production systems.

SUBSTANCE: system has first computer which can exchange data with second computer. Second computer transmits data to first computer. Data contains info on required color of ink but it doesn't obligatory has info on other required properties of ink. First computer has data base for predicting color data of ink receptions when using selected set of basic colors of ink, program for selecting receipt of ink based on data for ink required, and program for transmitting info to second computer for representing color which relates to selected receipt of ink. Second monitor has color display where indo is represented. Set of basic colors of ink can be selected on the base of other required properties of ink, for example, low cost, light resistance or chemical resistance.

EFFECT: close match of colors obtained.

15 cl, 3 dwg

The invention relates to information technology in medicine

The invention relates to automatic control and computer engineering and can be used to analyze the geometric parameters of the images of objects in computer vision systems

FIELD: physics, image processing.

SUBSTANCE: invention is related to digital processing of image in process of scanning and copying, and especially to the field of colour and black-and-white text segmentation, when text is automatically extracted from scanned document. According to the present invention, method of text segmentation by colour criterion consists in performance of the following operations: initial image is broken down into non-overlapping units of pixels; new image Z is generated, in which each pixel represents corresponding unit of initial image; in process of scanning serial units are selected from initial image; classification is carried out for current unit by criterion "monochromatic/colour" in space of opposite colours; initial colour unit RGB is broken down into monochromatic units R, G and B; detector of Laplacian-Gaussian edges is applied with specified threshold T to monochromatic unit; number of edges is calculated for each pixel in Z; classification of "text/non-text unit" is performed by comparison of edge number with set threshold C; classified channels are combined, using logical operator OR.

EFFECT: provision of universal approach to preliminary processing of initial document, providing for faster printing of copy and saving of multifunctional device resources.

4 cl, 3 dwg

FIELD: physics.

SUBSTANCE: invention relates to location of part of a document captured in an image. A method is proposed for determining part of a document, which corresponds to the captured image, when the user makes a flourish in the document. Location of some of the images will be determined, for example by analysing pattern on the capture image, or from results of pixel-by-pixel comparison of the image with the document. Location of other images will be determined through segmentation of the sequence of images in a group of corresponding forms of the flourish. Information relating to localised images in the segment can then be used for determining position of non-localised images in the segment. The search area of a non-localised image in the document is established based on the position of the previous localised image and maximum speed or maximum acceleration of the pen. Turning and scale of the non-localised image are evaluated, as well as turning and scale of the localised image, and the non-localised image is deformed using this turning and scale. Pixel-by-pixel comparison between the deformed non-localised image and the search area in the document is done. If deformation of the non-localised image is successfully matched, parametres of image transformation are further specified.

EFFECT: increased efficiency of locating part of a document.

10 cl, 25 dwg

FIELD: physics; computer engineering.

SUBSTANCE: invention relates to computer engineering and can be used for detecting image contours and subsequent input of the result into a computer. The device has a bus controller, RAM for reading (recording) intermediate image processing results, image filter unit, a contour detecting unit and a microprocessor.

EFFECT: more accurate detection of image contours.

1 dwg

FIELD: information technologies.

SUBSTANCE: method is proposed to identify contours of moving objects, including detection of moving pixels according to a principle of inter-frame subtraction, detection of direction of movement of detected pixels with account of direction of movement of adjacent pixels, generation of moving objects by combination of adjacent pixels with one direction and availability of pixels in an eight-connected vicinity, spatial differentiation of detected objects by a Sobel operator, skeletonisation of contour lines on a gradient image by the method of suppression of points of non-maximum brightness, threshold processing of a skeleton gradient image on the basis of maximum and minimum of enthropy.

EFFECT: higher accuracy in detection of areas of moving objects and higher speed for identification of image contours.

1 dwg

FIELD: information technology.

SUBSTANCE: fragments of a digital image similar top human faces are detected in all regions of the digital image; the indicator of the fraction of pixels of skin colours inside the region of a fragment is determined on said fragments; a face is detected if a given threshold value is exceeded; if the threshold value of the fraction of skin colour inside the region of the found face is exceeded, the initial colour palette of human skin is adjusted; outlines of the object of the colour skin are determined; an indicator of geometric similarity of the found outline with typical body outlines is calculated; fragments similar to typical naked body parts are searched; the image is analysed for pornographic visual content.

EFFECT: high accuracy of analysing content of a digital image.

16 cl, 2 dwg

FIELD: information technology.

SUBSTANCE: during training, a random number N with length of n bits is obtained; n artificial neurons are created; neuron inputs are randomly connected to memory cells with data on inclination of dermal ridges of fragments of examples of the classified fingerprint picture; further, the neurons are trained to output a random number N when presenting examples of the classified picture; the random number N is the hashed and the reference hash value is stored, as well as the connection of the neural network and its training parameters; and during authentication, there is repeated breaking up of the scanned picture into fragments, calculation of the average inclination on the fragment and the calculated value is transmitted to inputs of the neural network; the hash result is calculated at the output of the neural network and then compared with the reference value; in case of a match, personal authentication is continued, otherwise the authentication procedure is stopped once the time allocated for an authentication session runs out.

EFFECT: safeguarding the finger print picture and its biometric parameters which are used during biometric authentication, and providing conditions wherein it is impossible to find the owner of biometric parameters from biometric authentication data, while also cutting the average time of loading the authentication server.

2 cl, 3 dwg

FIELD: physics.

SUBSTANCE: system (200) has breaking unit (210) for breaking an image into a plurality of image areas, each image area displaying an area of the anatomical structure, and an adaptation unit (220) for adaptation of the referenced image to the evaluation function-based image, wherein the evaluation function is a function of parameters of the reference image and index numbers of the image and relative positions thereof on the image, and based on a criterion which must be satisfied by the index numbers of the evaluation function, wherein the evaluation function is determined based a plurality of image areas. Determination of the evaluation function based on a plurality of image areas enables to calculate optimum values of input into the value of the evaluation function within each image area.

EFFECT: high reliability of automatic detection of a cardiac muscle in short axis layer images.

11 cl, 9 dwg

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to digital image processing means. In the method, said operator is formed from linear structure-forming elements with different orientation parameters relative the raster of an image of equal length; each filtered image is obtained through interaction of the linear structure-forming element of the composite morphologic operator with the original image; pixel brightness in the filtered image is obtained by performing, for each pixel of the original image, three morphologic operations of interaction of the original image with the linear structure-forming element.

EFFECT: high accuracy of highlighting boundaries of complex-structure images owing to formation of multiple direction-filtered images from an original half-tone image via local processing with a composite morphologic operator.

6 dwg

FIELD: physics.

SUBSTANCE: disclosed is an apparatus for determining outlines of objects on an image. The apparatus comprises an image element sensor and the following units for: generating threshold values of speed and acceleration of change in intensity of the image elements, creating an eight-connected neighbourhood of the image elements, successive selection of triads in the eight-connected neighbourhood of an image element, memory for said triads, differentiating triads and a decision unit. The unit for generating threshold values of speed and acceleration of change in intensity of the image elements is connected to the output of the image element sensor. The unit for creating an eight-connected neighbourhood of the image elements is connected to the output of the unit for generating image elements and is connected to series-connected unit for successive selection of triads in the eight-connected neighbourhood of an image element, memory unit for triads of the eight-connected neighbourhood of an image element, triad differentiation unit and decision unit.

EFFECT: high noise-immunity of recognising an image outline by estimating the speed and acceleration of change of colour intensity in the eight-connected neighbourhood of an image element.

7 dwg

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to image forming apparatus with an identity authentication function. Disclosed is an image forming apparatus which includes an image forming unit. The apparatus also includes a recording unit, configured to record a plurality of data on characteristic features, up to the maximum limit, in non-volatile memory with reference to the same object. The apparatus also includes an authentication unit, configured to identify whether an object detected based on an image captured by the image forming unit is a recorded object based on data on characteristic features recorded in the nonvolatile memory. Furthermore, the apparatus includes a selection unit, configured to select one of the objects recorded in the nonvolatile memory, for which new data on characteristic features were recorded through the recording unit.

EFFECT: high accuracy of authenticating an object by selecting new data on characteristic features.

18 cl, 22 dwg

FIELD: printing ink production systems.

SUBSTANCE: system has first computer which can exchange data with second computer. Second computer transmits data to first computer. Data contains info on required color of ink but it doesn't obligatory has info on other required properties of ink. First computer has data base for predicting color data of ink receptions when using selected set of basic colors of ink, program for selecting receipt of ink based on data for ink required, and program for transmitting info to second computer for representing color which relates to selected receipt of ink. Second monitor has color display where indo is represented. Set of basic colors of ink can be selected on the base of other required properties of ink, for example, low cost, light resistance or chemical resistance.

EFFECT: close match of colors obtained.

15 cl, 3 dwg

FIELD: office equipment, in particular, system and method for correction of image during output to printing device.

SUBSTANCE: computer, with printing device connected to it, is equipped with information carrier, containing file with image, agent module, data storage module, at least one application, graphic device interface and printing device interface. Agent module is made with possible finding of new file with image on data carrier, aforementioned file containing additional information, calculations for this file of numeric index based on image data and recording of computed index together with additional file information to data storage module. Printing device driver is made with possible calculation of numeric index on basis of additional information data received in it, appropriate to calculated numeric index, and correction of image data received by it and transfer of corrected image data to printing device.

EFFECT: possible correction of image in accordance to additional information during output to printing device from various applications, installed on computer.

2 cl, 1 dwg

FIELD: information technologies.

SUBSTANCE: in color correction scheme per each tint any image which is combination of three different chroma signals is splitted into multiple ranges by tint so that colour correction may be performed by range of each tint. To prevent leaving of uncorrected peripheral range area for colour correction, colour correction procedure is performed so that correction ranges are overlapped with each other.

EFFECT: creation of colour correction procedure performed to provide possibility of true correction of all tints in image without leaving any uncorrected range.

5 cl, 4 dwg

FIELD: physics, image processing.

SUBSTANCE: invention is related to digital processing of image in process of scanning and copying, and especially to the field of colour and black-and-white text segmentation, when text is automatically extracted from scanned document. According to the present invention, method of text segmentation by colour criterion consists in performance of the following operations: initial image is broken down into non-overlapping units of pixels; new image Z is generated, in which each pixel represents corresponding unit of initial image; in process of scanning serial units are selected from initial image; classification is carried out for current unit by criterion "monochromatic/colour" in space of opposite colours; initial colour unit RGB is broken down into monochromatic units R, G and B; detector of Laplacian-Gaussian edges is applied with specified threshold T to monochromatic unit; number of edges is calculated for each pixel in Z; classification of "text/non-text unit" is performed by comparison of edge number with set threshold C; classified channels are combined, using logical operator OR.

EFFECT: provision of universal approach to preliminary processing of initial document, providing for faster printing of copy and saving of multifunctional device resources.

4 cl, 3 dwg

FIELD: instrument making.

SUBSTANCE: invention is related to calibration of colours in chamber and/or display design. Result is achieved by the fact that digital image generated with the help of chamber and/or display design, is corrected with the help of data base formed from information of reference colours spectrum.

EFFECT: correction of colour defects in digital images.

9 cl, 2 dwg

Up!