KR101528604B1 - Method and Apparatus for Processing Image - Google Patents

Abstract

The present invention relates to an image processing apparatus and method, comprising the steps of: scanning at least one pixel line in an image composed of a plurality of pixels in a predetermined direction to extract a change point of the pixel for each pixel line; A first distance between a changing point of the first pixel of the line and a changing point of the second pixel of the second pixel line located at one side of the first pixel line and a changing point of the first pixel, And determining a finishing point of the scan by comparing a second distance between the changing points of the third pixel of the third pixel line and a predetermined threshold.

Description

TECHNICAL FIELD The present invention relates to an image processing method and apparatus,

The present invention relates to an image processing method and apparatus, and more particularly, to an image processing method and apparatus capable of extracting a shape of a target object by removing noise around an object in the image.

In recent years, due to the development of digital devices, digital cameras or mobile phones can be used to easily capture images at anytime and anywhere. Digital devices capable of capturing an image such as a digital camera or a mobile phone have an image sensor for capturing an image therein.

However, when noise is included in the image data photographed by the image sensor, if the photographed image data is displayed, the image can not be output correctly. In addition, when the information about the shape of the object or the image data includes noise, it is difficult to accurately grasp the shape of the object due to the influence of the noise.

Accordingly, research is currently under way to more accurately reduce the noise and output the image in the image data. In particular, in order to minimize ambient noise other than the shape of the object in the image data and to extract the contour of the object Research is underway.

However, in the case of image data photographed in a general environment, when the contour of the object is extracted from the surrounding image of the object, that is, the background is always present in the background, the shape of the object due to the influence of noise in the background It was almost impossible to grasp correctly or there was a problem that was quite difficult.

KR 1996-0003296 A

Accordingly, an object of the present invention is to divide an image into a plurality of slices on the basis of a target object inside an input image, combine the contours of the object extracted from each of the plurality of fragments, And to provide an image processing method and apparatus capable of extracting images.

According to an aspect of the present invention, there is provided an image processing method comprising: scanning at least one pixel line in an image composed of a plurality of pixels in a predetermined direction to extract a change point of the pixel for each pixel line; And a first distance between a change point of the first pixel of the first pixel line and a change point of the second pixel of the second pixel line located at one side of the first pixel line, And comparing the second distance between the change points of the third pixel of the third pixel line located on the other side of the first pixel line with a predetermined threshold to determine the end point of the scan.

The step of determining the end point of the scan may include continuing the scan to extract another pixel change point of the first pixel line if both the first and second distances are greater than the threshold .

The method of claim 1, wherein determining the end of the scan comprises: if the first and second distances are greater than the threshold and the other is less than the threshold, Grouping may be performed in a direction of a pixel line in which a distance between both sides is smaller than the threshold value and the other is larger than the threshold value.

Wherein if the number of lines of the grouped pixel lines is smaller than a predetermined number of lines, the scanning is continued to extract another pixel change point with respect to the grouped pixel lines, and if the number of lines of the grouped pixel lines is smaller than the predetermined number of lines And terminating the scan if the size is larger.

The ending point of the scan may include terminating the scan if the first and second distances are both less than the threshold.

The step of extracting the change point of the pixel may include extracting a pixel whose value of the pixel is larger than a predetermined reference value as a change point of the pixel.

Meanwhile, the image processing apparatus according to an embodiment of the present invention scans at least one pixel line in an image made up of a plurality of pixels along a predetermined direction to extract change points of the pixels for each pixel line, A first distance between a changing point of the first pixel of the line and a changing point of the second pixel of the second pixel line located at one side of the first pixel line and a changing point of the first pixel, And an extracting unit for comparing the second distance between the change points of the third pixel of the third pixel line and a predetermined threshold to determine the end point of the scan.

The extraction unit may continue the scan to extract another pixel change point of the first pixel line if both the first and second distances are greater than the threshold value.

The extraction unit groups the pixels in the direction of a pixel line having a distance smaller than the threshold value from the first pixel line if any one of the first and second distances is larger than the threshold and the other is smaller than the threshold, Can be grouped up to a pixel line where one side is smaller than the threshold value and the other side is larger than the threshold value.

Wherein the extracting unit continues the scan so as to extract another pixel change point for the grouped pixel line if the number of lines of the grouped pixel line is smaller than a predetermined number of lines, If the number of lines is larger than the predetermined number of lines, the scan can be terminated.

The extractor may terminate the scan if the first and second distances are all less than the threshold.

The extracting unit may extract a pixel having a pixel value larger than a predetermined reference value as a change point of the pixel.

According to the image processing method and apparatus according to the embodiment of the present invention, an image is divided into a plurality of fragments on the basis of a target object in the input image, and contours of the target extracted from each of the plurality of fragments are combined, There is an advantage that the final contour line can be extracted more accurately.

That is, there is an advantage that the shape of the object can be grasped correctly by removing the noise around the object.

1 is a configuration diagram of an image processing apparatus according to an embodiment of the present invention.
2 is an example of an input image.
FIG. 3 shows an outline of a subject extracted roughly using a Sobel algorithm in an input image.
4 is an example of the upper and lower images divided into a plurality of fragments.
Fig. 5 is an example showing contours of a target object extracted from a plurality of fragments on the upper side.
6 shows an example of Left sweeping in which an image is left-scanned.
FIG. 7 shows an example of top sweeping in which an image is top-scanned.
FIGS. 8 and 9 are examples of a top sweeping process for extracting contours of a target object in an image in which the Sobel algorithm is performed.
10 is an example showing sweeping using a data pattern.
11 shows an example of threshold sweeping using the distance between simple sweeping results.
FIGS. 12 and 13 show an example of a process of extracting contours of a target object from fragments # 8 through # 13 according to an embodiment of the present invention.
FIG. 14 is an example showing contours of extracted objects through an iterative contour extraction process.
Fig. 15 is an example showing an image in which the noise around the object is removed.
16 is an example showing the final contour of the object.
17 is a flowchart illustrating an image processing process according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. Some of the drawings are shown in reversed processing so that the expressions can be made clear.

FIG. 1 is a configuration diagram of an image processing apparatus according to an embodiment of the present invention. FIG. 2 shows an example of an input image. FIG. 3 shows an outline of a target object extracted roughly using a Sobel algorithm in an input image. Fig. 4 shows an example of the upper and lower images divided into a plurality of fragments, and Fig. 5 shows an example showing the contours of the object extracted from the plurality of fragments on the upper side.

As shown in FIG. 1, the image processing apparatus 100 includes a preprocessing unit 110, a dividing unit 120, an extracting unit 130, a combining unit 140, and a post-processing unit 150.

As shown in FIG. 2, the preprocessing unit 110 may perform a preprocessing operation on an image input from a digital device including a digital camera or a mobile phone. That is, the preprocessing unit 110 can reduce the noise in the input image through the preprocessing process. Particularly, the preprocessing unit 110 uses an algorithm for eliminating motion blur and an algorithm for sharpening the edge of the image Thereby reducing overall noise in the image as needed.

Also, as shown in FIG. 3, the preprocessing unit 110 can roughly extract the contour of the object in the noise-reduced image in the preprocessing process. The preprocessing unit 110 may use a Sobel algorithm or a Prewitt algorithm to roughly extract the contour of the object during the preprocessing process. The Sobel algorithm is the most representative first-order differential operator capable of extracting a contour (edge) that is a discontinuity point where the brightness of a pixel changes suddenly at the boundary between an object (object) and a background. It is advantageous to extract edges in the diagonal direction, and there is generally a strong advantage in noise. Prewitt algorithm is mainly used for extracting vertical and horizontal edges rather than diagonal direction, and has a high speed of execution. It goes without saying that the above-described preprocessing process can be omitted according to the set conditions.

On the other hand, the object appearing in the image must satisfy the following conditions.

First, the shape of the object should all appear in the image. Second, the object should be located in the center of the image. Third, the parts corresponding to both end points of the object should be at the center line of the image as much as possible. In order to satisfy the condition of the object, it is possible to perform the operation of adjusting the position of the object in the image after performing the preprocessing process or the preprocessing process.

As shown in FIG. 4, the dividing unit 120 divides the image into upper and lower images based on both end points of the object in the input image, divides the upper and lower images into a plurality of slices have. Here, the upper and lower images are referred to as top slice and bottom slice, respectively, and the upper and lower halves can be divided into the same number (for example, 20 in FIG. 4).

5, the extraction unit 130 extracts contours of a target object from a plurality of fragments constituting the upper fragment, and then extracts contours of a plurality of fragments constituting the lower fragment The outline of the object can be extracted from the fragment.

More specifically, the extracting unit 130 scans at least one pixel line, which is a fragment, along a predetermined direction to search for a change point of a pixel for each pixel line, The outline of the object can be extracted. That is, the extracting unit 130 may scan the plurality of pixel lines in a predetermined direction by using a sweeping algorithm to find a pixel (data) dividing point indicating a shape or a color in each pixel line , And the contour of the object (or the turning point of the brightness or the color) of the object can be extracted by applying this.

Hereinafter, the sweeping algorithm used to extract the contour of the object will be described in more detail.

First, sweeping algorithms can be classified as follows: 1) a simple sweeping algorithm using linear scans, 2) a pattern sweeping algorithm using data patterns, and 3) a threshold sweeping algorithm using the distance between simple sweeping results ≪ / RTI >

Among them, 1) a simple sweeping algorithm using a linear scan, when a pixel (data) satisfying a predetermined condition is found during a linear scan along a plurality of pixel lines forming a fragment, An algorithm of selecting a pixel found as a change point of the pixel by performing a scan continuously to the end of a row or a column and may be performed on the basis of a linear scan.

A linear scan is a linear scan of each of a plurality of pixel lines, and the direction of the linear scan may be bi-directional with respect to one axis. For example, in a two-dimensional (2D) image, a linear scan can be performed in four scan directions: Top scan (scan from top to bottom), Bottom scan (scan from bottom to top), Right scan ) And Left scan (scan from left to right), and the sweep algorithm can be divided into Top Sweep, Bottom Sweep, Right Sweep, and Left Sweep according to the scan direction. Here, the linear scan can be arbitrarily set according to selection conditions such as searching for specific data, difference in data, a range of specific data or a value included in a set, and the end point of the linear scan can be determined arbitrarily Of course.

FIG. 6 shows an example showing left sweeping in which an image is left-scanned, and FIG. 7 shows an example showing top sweeping in which an image is top-scanned.

Referring to FIG. 6, the left sweep algorithm scans each of the twelve pixel lines from left to right to search for a pixel having a value of 5 or more in each pixel line, It is possible to extract the change point of the pixel by stopping the scan of the row or the column. FIG. 7 is a diagram illustrating a method of scanning each of the twelve pixel lines in the downward direction under the same conditions as in FIG. 6, searching for a pixel having a value of 5 or more in each pixel line, If found, the scan of the row or column is stopped, and the change point of the pixel can be extracted.

FIGS. 8 and 9 are examples of a top sweeping process for extracting contours of a target object in an image in which the Sobel algorithm is performed.

As shown in FIG. 8, when the Sobel algorithm is performed and top sweeping is performed on the image in which the contour of the object is roughly extracted, the contour corresponding to the upper part of the object shown in FIG. 9 can be extracted.

Next, 2) a pattern sweeping algorithm using a data pattern is an algorithm for finding an array of two or more specific pieces of information. For example, finding a pair such as 'black-white-black' do.

FIG. 10 shows an example of sweeping using a data pattern. In FIG. 6, an image composed of 12 pixel lines shown in FIG. 6 is subjected to Left pattern scan (pattern scan from left to right) If a pattern of '0-8' is found during the pattern scan, the scan is stopped immediately upon detection, and the result of pattern sweeping as shown in FIG. 10 can be extracted.

As described above, the extracting unit 130 may search for a pixel satisfying a predetermined condition in each of a plurality of pixel lines by using a simple sweeping algorithm using a linear scan or a pattern sweeping algorithm using a data pattern And combines the retrieved pixels to extract the contour of the object.

The extraction unit 130 scans at least one pixel line constituting the fragment in accordance with a predetermined direction to search for a change point of the pixel for each pixel line and determines whether the change point of the pixel or both sides of the change point of the pixel The edge of the pixel can be extracted by comparing the distance between the pixel change points on the pixel line located with the predetermined threshold value to determine the end point of the scan and further searching for the change point of the pixel according to the end point of the determined scan.

More specifically, the extractor 130 may extract a contour of a target object using a threshold sweep algorithm using a distance between simple sweep results (hereinafter referred to as a 'threshold sweep algorithm'), where , And the threshold sweep algorithm is an algorithm for determining whether to terminate the linear scan on each pixel line using the result of simple sweeping using linear scan.

11 shows an example of threshold sweeping using the distance between simple sweeping results.

Referring to FIG. 11, there are a plurality of scan lines (pixel lines) and the change points of the pixels found in the scan lines' n ',' n-1 'and' n + When each is assumed that S (n-1) and S (n + 1), represents the distance between S (n) and S (n-1) with DELTA _{n (n-1), S} (n) and S (n +1) can be expressed as ㅿ _n (n + 1). If both of _{n n} (n-1) and _{n n} (n + 1) are greater than a predetermined threshold value D max, scanning continues on the scan line 'n' to continue extracting the change point of the pixel, _{If both n} (n-1) and _n (n + 1) are not greater than the threshold value Dmax, the scan at the scan line 'n' is terminated and the operation of extracting the change point of the pixel can be stopped. If only one of the distances of _nn (n-1) and _nn (n + 1) is larger than the threshold value Dmax, the scan in the scan line 'n' The scan can continue.

When the scan for the entire scan line is completed, it is determined whether or not there is a temporarily terminated scan line. If the scan line is temporarily terminated, the next scan end line is shifted toward the scan line at a distance smaller than the threshold value Dmax, The scan lines can be grouped up to the scan line of FIG. For example, DELTA _n (n-1) is larger than the threshold value (Dmax), DELTA _{n (n} + 1), DELTA _{n + 1 (n +} 2), and DELTA _{n + 2 (n +} 3) is the threshold value (Dmax ) smaller, DELTA _{n + 3} (n + 4) side is larger than the threshold value (Dmax), n and (n + 3) th scanning line is a scanning line of the temporary termination state, and thus n to (n + 3) th The scan lines may be grouped into a group, and they are spaced apart from the scan lines before and after the scan line group.

In the case of a scan line group in which the number of lines is smaller than the predetermined number of lines in the grouped scan lines, the scan lines included in the group are further scanned to find the next pixel change point. In the case of a scan line group in which the number of lines is larger than the predetermined number of lines, the temporary end state can be released and the scan can be terminated.

As described above, the extracting unit 130 searches for a change point of a pixel for each pixel line using a threshold sweeping algorithm using the distance between simple sweeping results, and detects a change point of the pixel and a pixel line And determines the end point of the scan by searching the change point of the pixel according to the end point of the determined scan to extract the contour of the object.

More specifically, the combining unit 140 may calculate a standard deviation for each of a plurality of fragments from which the outlines are extracted, and calculate a plurality of fragments using the calculated standard deviation, Can be combined. That is, the combining unit 140 may calculate the standard deviation of the contour information in each fragment, and may combine a plurality of fragments using the standard deviation of the contour information, wherein the contour information is from the center line to the contour in the fragment (X, Y), or may be represented by coordinate information with respect to (X, Y).

In order to calculate the standard deviation of the contour information in each fragment, the combining unit 140 preferentially calculates the average information of the contour, which is the average of the contour information in each fragment, and the information of the contour in each fragment is the average By taking the distance away from the information and calculating the standard deviation, it is possible to combine multiple fragments based on this.

More specifically, the combining unit 140 selects a reference fragment having a minimum standard deviation among fragments having a standard deviation smaller than a predetermined allowable standard deviation, and sequentially selects the fragments located on both sides of the reference fragment on the basis of the selected reference fragment You can combine to create a combined image. The permissible standard deviation is the maximum allowable standard deviation. The combining unit 140 can select the reference segment having the smallest standard deviation among the segments having the standard deviation smaller than the allowable standard deviation, If there is a fragment with a smaller standard deviation, the reference fragment and the fragment may be combined into one fragment to generate a combined image. Here, a given image can be regarded as rejection or failure when there are no suitable fragments or the number of fragments within a specified standard deviation does not reach a certain limit.

In addition, the combining unit 140 performs curve-fitting on each fragment included in the combined image to calculate virtual contour information, and compares the original contour information and the virtual contour information in each fragment So that the difference between the original outline information and the virtual outline information can be detected. Then, the combining unit 140 uses the original outline information as it is for a section where the difference between the original outline information and the virtual outline information is smaller than or equal to a predetermined reference difference, and uses the detected original outline information and the virtual outline information The original outline information corresponding to the section in which the information difference is larger than the predetermined reference difference can be replaced with the virtual outline information. That is, the combining unit 140 can remove noise from original contour information extracted from each fragment by replacing the original contour information determined as noise with virtual contour information.

On the other hand, when the standard deviation of the fragment located on both sides of the reference fragment is not smaller than the allowable standard deviation, the combining unit 140 calculates virtual contour information for the first fragment whose standard deviation is not smaller than the allowable standard deviation, The contour information of the first fragment may be corrected according to the contour information of the contour information, and the combined first image and the combined contour information may be combined. Here, the virtual contour information can be calculated using the contour information and the preset polynomial for the fragment in the combined image.

More specifically, if there is no fragment having a standard deviation smaller than the allowable standard deviation among the fragments positioned adjacent to the reference fragment, the combining unit 140 detects the polynomial based on the contour information in the previously combined image, The detected polynomial can be used to perform curve-fitting on the first piece having a larger standard deviation than the allowable standard deviation. Here, curve fitting is one of the process of fitting a function to a given data set, and can be performed using a linear function, a polynomial function, or an exponential function. In other words, curve fitting is often used when estimating the value between given data when approximating data, and when fitting a data set by curve fitting using polynomial, the polynomial passes through all the data, Curve fitting can be done to approximate the data as a whole.

FIGS. 12 and 13 show an example of a process of extracting contours of a target object from fragments # 8 through # 13 according to an embodiment of the present invention.

12, it is assumed that the allowable standard deviation is 15. In the fragments 8 to 13, the standard deviation of the 8th fragment is 33.5, the fragment 9 is 26.3, the fragment 10 is 14.8, 11 can be calculated as 8.6, 12 as 12, and 13 as 5.1. In this case, the 13th fragment is the reference fragment with a standard deviation of 5.1 and has the highest reliability. Since the standard deviation of the 10th to 12th fragments is smaller than the allowable standard deviation, the 13th fragment is divided into 10th to 12th fragments Thereby creating a combined image. Then, the contour information from the 10th to 13th fragments is used as it is.

However, since the 9th piece adjacent to the combined image, which is the combined pieces, exceeds the allowable standard deviation of 15, the contour information of the 9th piece is deleted. Instead, the curve fitting is performed based on the contour information of the 10th, 11th, 12th, And the virtual contour information corresponding to the 9th fragment is calculated and combined using the detected polynomial. And the 8th fragment will be merged in the same way. The result of this example is shown in Fig. 13, in which the original data, outline information, is used for the fragments 10 to 13, but fragments 8 and 9 are modified to the virtual contour information using the polynomial Able to know.

14 illustrates an outline of a target object extracted through an iterative contour extraction process. Referring to FIG. 14, a contour of a target object as shown in FIG. 14 can be extracted by repeating the process of combining fragments based on the merged fragment. If there are no fragments less than the allowable standard deviation among the neighboring fragments, the curve fitting process is repeated, and the merge process is repeated based on the combined fragments. Can be performed until one is combined.

The post-processing unit 150 can perform a post-processing operation on the image of the object from which the outline has been extracted. More specifically, the post-processing unit 150 removes vertices having a predetermined interval from the outline of the sweeping image Band-pass filtering. Then, the post-processing unit 150 may repeat the operation of extracting the outline with the filtered image by the extracting unit 130. [ At this time, the number of iterations and the interval used in the band-pass can be adjusted according to the state of the image to be used or the characteristic of the object to be searched.

15 is an example showing an image in which noises around the object are removed, and FIG. 16 is an example showing a final contour line of the object. As shown in FIG. 15, the post- In the extracted image of the contour, the part outside the object, ie, the part outside the object, can be removed (Noise Removal).

In addition, the post-processing unit 150 may apply the extracting operation of the outline based on the outline image from which the noise is removed by the extracting unit 130. [ At this time, the above-described process can be repeated in whole or in part as needed, and it is also possible to extract the final contour line as shown in FIG. 16 by applying another algorithm. Here, the final outline may be completely closed according to the used image, for example, an outline to which all the outlines are connected, but the outline is mostly partially opened without being connected. At this time, if you need to close an open contour according to the intended use, the algorithm for this is different from noise removal and can use curve fitting or other algorithms.

Hereinafter, an image processing process according to an embodiment of the present invention will be described.

17 is a flowchart illustrating an image processing process according to an embodiment of the present invention.

As shown in FIG. 17, an image is divided into upper and lower portions based on both end points of a target object in the input image (S210), and the upper and lower divided images are divided into a plurality of slices (S220). Here, the upper and lower images are referred to as a top slice and a bottom slice, respectively, and the upper and lower halves can be divided into the same number of fragments.

In this way, before the image is divided into upper and lower parts, a preprocessing operation can be performed on an image input from a digital device including a digital camera or a mobile phone. In other words, it is possible to reduce the noise in the input image through the preprocessing process. In particular, by using an algorithm for eliminating the motion blur and an algorithm for sharpening the edge of the image, The overall noise can be reduced.

In addition, it is possible to roughly extract the contour of the object in the noise-reduced image in the preprocessing process. That is, the contour of the object can be roughly extracted during the preprocessing process using a Sobel algorithm or a Prewitt algorithm. The Sobel algorithm is the most representative first-order differential operator capable of extracting a contour (edge) that is a discontinuity point where the brightness of a pixel changes suddenly at the boundary between an object (object) and a background. It is advantageous to extract edges in the diagonal direction, and there is generally a strong advantage in noise. Prewitt algorithm is mainly used for extracting vertical and horizontal edges rather than diagonal direction, and has a high speed of execution. It goes without saying that the above-described preprocessing process can be omitted according to the set conditions.

On the other hand, the object appearing in the image must satisfy the following conditions.

First, the shape of the object should all appear in the image. Second, the object should be located in the center of the image. Third, the parts corresponding to both end points of the object should be at the center line of the image as much as possible. In order to satisfy the condition of the object, it is possible to perform the operation of adjusting the position of the object in the image after performing the preprocessing process or the preprocessing process.

Then, contours of the object are extracted from each of the plurality of fragments (S230). That is, the outline of the object can be extracted from a plurality of fragments constituting the upper fragment, and the outline of the object can be extracted from the plurality of fragments constituting the lower fragment.

More specifically, at least one pixel line constituting a fragment is scanned in a predetermined direction to search for a change point of a pixel for each pixel line, and the change point of the retrieved pixel is combined, Can be extracted. That is, the extracting unit 130 can scan a plurality of pixel lines in a predetermined direction by using a sweeping algorithm to find a dividing point of pixels (data) representing a shape or a color in each pixel line , And the contour of the object (or the turning point of the brightness or the color) of the object can be extracted by applying this.

In addition, at least one pixel line constituting a fragment is scanned according to a predetermined direction to search for a change point of a pixel for each pixel line, and a change point of a pixel and a pixel in a pixel line located on both sides of the change point of the pixel And determines the end point of the scan. Further, the contour of the object can be extracted by additionally searching the change point of the pixel according to the end point of the determined scan. That is, a contour of a target object can be extracted using a threshold sweeping algorithm using the distance between simple sweeping results.

Then, a plurality of fragments from which the outline is extracted are combined (S240).

More specifically, a standard deviation may be calculated for each of a plurality of fragments from which a contour is extracted, and a plurality of fragments may be combined using the calculated standard deviation. That is, it is possible to calculate the standard deviation of the contour information in each fragment, and combine a plurality of fragments using the standard deviation of the calculated contour information, wherein the contour information is the straight line distance from the center line to the contour in the fragment (X, Y), as shown in Fig.

In order to calculate the standard deviation of the contour information of each fragment, the average information of the contour, which is the average of the contour information in each fragment, is preferentially calculated, and the information of the contour in each fragment is calculated as the distance The standard deviation is calculated, and a plurality of fragments can be combined based on the standard deviation.

More specifically, a reference fragment having a minimum standard deviation is selected from fragments whose standard deviation is smaller than a predetermined allowable standard deviation, and fragments located on both sides of the reference fragment are combined based on the selected reference fragment to generate a combined image . The allowable standard deviation is the maximum allowable standard deviation. A fraction with the smallest standard deviation among the fractions whose standard deviation is smaller than the allowable standard deviation can be selected and a standard deviation smaller than the allowable standard deviation on both sides of the selected reference fragment The combined image can be generated by combining the reference fragment and the fragment into one fragment. Here, a given image can be regarded as rejection or failure when there are no suitable fragments or the number of fragments within a specified standard deviation does not reach a certain limit.

Further, curve-fitting is performed on each fragment in the combined image to calculate virtual contour information, and the original contour information in each fragment is compared with the virtual contour information to obtain the original contour information and virtual It is possible to detect the difference of the outline information of the image. When the difference between the original contour information and the virtual contour information is smaller than or equal to a predetermined reference difference, the original contour information is used as it is, and the difference between the detected original contour information and the virtual contour information is determined And for the section longer than the reference difference, the original contour information corresponding to the section can be replaced with the virtual contour information. That is, from the original outline information extracted from each fragment, original outline information determined as noise can be replaced with virtual outline information to remove noise.

If the standard deviation of the fragment located on both sides of the reference fragment is not smaller than the allowable standard deviation, the virtual contour information is calculated for the first fragment whose standard deviation is not smaller than the allowable standard deviation, 1 piece of contour information, and combine the combined first image and the combined image with the modified contour information. Here, the virtual contour information can be calculated using the contour information and the preset polynomial for the fragment in the combined image.

More specifically, if there is no fragment having a standard deviation smaller than the allowable standard deviation among the fragments positioned adjacent to the reference fragment, the polynomial is detected based on the contour information in the previously combined image and the detected polynomial is used It is possible to perform curve-fitting on the first piece having a standard deviation larger than the allowable standard deviation. Here, curve fitting is a process of fitting a function to a given data set, and curve fitting can be performed using a linear function, a polynomial function, an exponential function, or the like. In other words, curve fitting is often used when estimating the value between given data when approximating data, and when fitting a data set by curve fitting using polynomial, the polynomial passes through all the data, Curve fitting can be done to approximate the data as a whole.

Curve fitting is a method of calculating and calculating a curve suitable for scattered points on a plane, for example, sampling a certain waveform and obtaining data. For this sampled data, the function of the original waveform curve can be calculated and the appropriate one can be displayed. The sum of squared error between the function and each measured value is obtained, and a coefficient is given to each variable so that it is minimized. Method, the deviation of the curve from each point measured by this method can be minimized.

Thereafter, a post-processing operation can be performed on the image of the object from which the outline has been extracted. In more detail, an operation of extracting the final outline of the object by removing noise from the image of the extracted object can be performed .

Embodiments of the present invention include a computer-readable medium having program instructions for performing various computer-implemented operations. This medium records a program for executing the above-described image processing method. The medium may include program instructions, data files, data structures, etc., alone or in combination. Examples of such media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD and DVD, programmed instructions such as floptical disk and magneto-optical media, ROM, RAM, And a hardware device configured to store and execute the program. Or such medium may be a transmission medium, such as optical or metal lines, waveguides, etc., including a carrier wave that transmits a signal specifying a program command, data structure, or the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, Of the right.

100: Image processing device
110:
120: minute installment
130:
140:
150: Post-

Claims (12)

Scanning at least one pixel line in an image made up of a plurality of pixels along a predetermined direction to extract a change point of the pixel for each pixel line,
A first distance between a changing point of the first pixel of the first pixel line and a changing point of the second pixel of the second pixel line located at one side of the first pixel line and a changing point of the first pixel, Determining a finishing point of the scan by comparing a second distance between points of change of a third pixel of a third pixel line located on the other side of the first pixel and a predetermined threshold,
Determining a change point of the first pixel as a last change point of the first pixel line when the end of the scan is determined and continuing to scan the first pixel line if the end of the scan is not determined
/ RTI > The method of claim 1,
Wherein determining the end of the scan comprises:
And if the first and second distances are both greater than the threshold, proceeding with the scan to extract another pixel change point of the first pixel line. The method of claim 1,
Wherein determining the end of the scan comprises:
If one of the first and second distances is larger than the threshold value and the other is smaller than the threshold value, grouping is performed in a pixel line direction having a distance smaller than the threshold value from the first pixel line, And performing grouping up to a pixel line that is smaller than the threshold and the other is larger than the threshold. 4. The method of claim 3,
Wherein if the number of lines of the grouped pixel lines is smaller than a predetermined number of lines, the scanning is continued to extract another pixel change point with respect to the grouped pixel lines, and if the number of lines of the grouped pixel lines is smaller than the predetermined number of lines And terminating the scan if the size of the image is larger. The method of claim 1,
Wherein determining the end of the scan comprises:
And terminating the scan if both the first and second distances are less than the threshold. The method of claim 1,
Wherein the step of extracting the change point of the pixel comprises:
And extracting a pixel whose value of the pixel is larger than a predetermined reference value as a change point of the pixel. The method comprising the steps of: scanning at least one pixel line in an image of a plurality of pixels along a predetermined direction to extract a change point of the pixel for each pixel line; Between a change point of the first pixel and a change point of the third pixel of the third pixel line located on the other side of the first pixel line, a first distance between a change point of the second pixel of the second pixel line located at one side of the first pixel line, Determining a change point of the first pixel as a final change point of the first pixel line when the end of the scan is determined; Is not determined, the extraction unit
And the image processing apparatus. 8. The method of claim 7,
The extracting unit extracts,
And if the first and second distances are both greater than the threshold, continue scanning to extract another pixel change point of the first pixel line. 8. The method of claim 7,
The extracting unit extracts,
If one of the first and second distances is larger than the threshold value and the other is smaller than the threshold value, grouping is performed in a pixel line direction having a distance smaller than the threshold value from the first pixel line, Wherein the grouping is performed up to a pixel line that is smaller than the threshold value and the other of which is larger than the threshold value. The method of claim 9,
The extracting unit extracts,
Wherein if the number of lines of the grouped pixel lines is smaller than a predetermined number of lines, the scanning is continued to extract another pixel change point with respect to the grouped pixel lines, and if the number of lines of the grouped pixel lines is smaller than the predetermined number of lines And terminates the scanning if the size of the image is larger than the size of the image. 8. The method of claim 7,
The extracting unit extracts,
And terminates the scan if both the first and second distances are less than the threshold. 8. The method of claim 7,
The extracting unit extracts,
And extracts, as a change point of the pixel, a pixel whose value of the pixel is larger than a predetermined reference value.

Images