JP2016035670A - Information processor, blurring level calculation method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve determination accuracy of blurring level.SOLUTION: There is provided an information processor 10 which includes: a storage section 11 that stores a piece of related information which represents a relation between a blurring index as an index of blurring level of image and the blurring level according to the width of a line in a character included in an image; and a calculation unit 12 that calculates a line width and a blurring index using an image and makes a reference to the relation information stored in the storage section 11 to identify the blurring level corresponding to the line width and the blurring index. The information processor 10 determines the blurring level with a high accuracy even when a line of a character is deformed etc.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an information processing apparatus, a blur condition calculation method, and a program.

  In recent years, character recognition technology that automatically recognizes characters from input images in various scenes has been used. For example, a character recognition technique is used in an information providing system that recognizes characters from landmark images captured by a camera and provides information according to the recognized characters. As a character recognition method, for example, there is a method in which a dictionary image representing a set character and an input image are compared in pixel units to calculate a similarity, and a dictionary image with the highest similarity is used as a recognition result.

  When the above method is applied, if the input image is blurred, the difference between the dictionary image and the input image becomes large, and the accuracy of character recognition decreases. For example, small characters and characters having complicated shapes such as kanji are easily affected by blurring of the input image. Under such circumstances, a method has been proposed in which a plurality of dictionary images having different degrees of blur are prepared for the same character, and a dictionary image having a high degree of similarity is extracted by comparing the input image with each dictionary image. When this method is applied, images that are blurred in the same manner are compared with each other, which may improve the accuracy of character recognition.

  However, when all of the plurality of dictionary images prepared for each character are compared with the input image, the amount of calculation for the comparison process increases, and the processing time for character recognition increases. On the contrary, if the comparison process is executed on a dictionary image whose degree of blur is close to that of the input image, the processing time can be shortened.

  A method has been proposed in which the degree of deterioration of a character to be recognized is specified and a character recognition method is selected according to the specified degree of deterioration. In this method, a deterioration index is obtained from an input pattern, and a dictionary matching the deterioration index is selected. Further, a method has been proposed in which character recognition is performed using density features when the degree of deterioration is large, and character recognition is performed using direction features when the degree of deterioration is small.

JP 2007-304899 A JP-A-11-175562 JP 2007-026027 A

  When the above comparison process is performed using a dictionary image that is close to the degree of blurring of the input image, the accuracy of character recognition decreases if a shift occurs in the degree of blurring between the input image and the dictionary image. The degree of blur is calculated based on, for example, the luminance gradient and the edge width at the edge portion of the character. The details of characters included in a blurred image may be crushed. When collapse or the like occurs, there is a possibility that an error occurs in the detection of the luminance gradient or the edge width. If an error occurs in the detection of the brightness gradient or the edge width, the accuracy of determining the degree of blurring may be reduced, and the accuracy of character recognition may be reduced.

  Therefore, according to one aspect, an object of the present invention is to provide an information processing apparatus, a blur condition calculation method, and a program that can improve the determination accuracy of the blur condition.

  According to one aspect of the present disclosure, a storage unit that stores relationship information indicating a relationship between a blur index that is an index of a blur condition of an image and a blur condition according to a line width of a character included in the image, and a captured image An information processing apparatus is provided that includes a calculation unit that calculates a line width and a blur index from the reference unit, and refers to the relationship information stored in the storage unit, and specifies a degree of blur corresponding to the line width and the blur index. The

  Further, according to another aspect of the present disclosure, the computer stores relationship information indicating a relationship between a blur index that is an index of a blur condition of an image and a blur condition according to a line width of a character included in the image. The relationship information is acquired from the storage unit, the line width and the blur index are calculated from the captured image, and the blur condition corresponding to the line width and the blur index is specified with reference to the relationship information stored in the storage unit A method for calculating the degree of blur is provided.

  According to another aspect of the present disclosure, the computer stores relation information indicating a relationship between a blur index that is an index of a blur condition of an image and a blur condition according to a line width of a character included in the image. The relationship information is acquired from the storage unit, the line width and the blur index are calculated from the captured image, and the blur condition corresponding to the line width and the blur index is specified with reference to the relationship information stored in the storage unit A program for executing processing is provided.

  According to the present invention, it is possible to improve the determination accuracy of the degree of blur.

It is the figure which showed an example of the information processing apparatus which concerns on 1st Embodiment. It is the figure which showed an example of the character recognition system which concerns on 2nd Embodiment. It is a 1st figure for demonstrating the character recognition system which concerns on 2nd Embodiment. It is a 2nd figure for demonstrating the character recognition system which concerns on 2nd Embodiment. It is a figure for demonstrating the relationship between a blur condition, edge gradient, and edge width. It is a figure which shows the brightness | luminance of an edge part. It is a figure which shows edge width. It is a figure for demonstrating normalization. It is a figure for demonstrating normalization with respect to the image where the line was crushed. It is a figure which shows the relationship between an edge gradient and a circle of confusion. It is a figure which shows the relationship between the width | variety of a line, an edge gradient, and a circle of confusion. It is a figure which shows an example of the area | region used for calculation of line | wire width. It is a figure for demonstrating calculation of line | wire width. It is the block diagram which showed an example of the function which the image processing apparatus which concerns on 2nd Embodiment has. It is the figure which showed an example of the confusion circle specific information which concerns on 2nd Embodiment. It is the figure which showed an example of the dictionary information which concerns on 2nd Embodiment. It is a flowchart which shows operation | movement of a confusion circle specific information selection part. It is the figure which showed an example of the hardware which can implement | achieve the function of the image processing apparatus which concerns on 2nd Embodiment. 6 is a flowchart illustrating an operation of the image processing apparatus according to the second embodiment.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. In addition, about the element which has the substantially same function in this specification and drawing, duplication description may be abbreviate | omitted by attaching | subjecting the same code | symbol.

<1. First Embodiment>
The first embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of an information processing apparatus according to the first embodiment. Moreover, the information processing apparatus 10 illustrated in FIG. 1 is an example of an information processing apparatus according to the first embodiment.

The information processing apparatus 10 includes a storage unit 11 and a calculation unit 12.
The storage unit 11 is a volatile storage device such as a RAM (Random Access Memory) or a non-volatile storage device such as an HDD (Hard Disk Drive) or a flash memory. The arithmetic unit 12 is a processor such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). However, the arithmetic unit 12 may be an electronic circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). The computing unit 12 executes a program stored in the storage unit 11 or another memory, for example.

  The storage unit 11 stores relation information indicating a relationship between a blur index that is an index of a blur condition of an image and a blur condition according to a line width of a character included in the image. The calculation unit 12 calculates the line width and the blur index from the captured image, and refers to the relationship information stored in the storage unit 11 to identify the degree of blur corresponding to the line width and the blur index.

  FIG. 1 shows, as an example, function curves F1, F2, and F3 corresponding to a plurality of line widths Wa, Wb, and Wc, respectively. The function curves F1, F2, and F3 are examples of relationship information. However, in the example of FIG. 1, the relationship information is expressed by a continuous curve function, but the relationship information can also be expressed as information in a table format that associates the degree of blur with the blur index. Note that the blur index is, for example, the gradient of the edge portion or the width of the edge portion.

  For example, it is assumed that the line width of the character calculated by the calculation unit 12 from the captured image is Wb. It is assumed that the blur index calculated by the calculation unit 12 is y. In this case, the calculation unit 12 selects the function curve F2 corresponding to the line width Wb from the function curves F1, F2, and F3. And the calculating part 12 specifies the blur condition x corresponding to the blur parameter | index y using the function curve F2.

  As described above, the relationship information indicating the relationship between the blur index and the blur condition is prepared in advance, and the blur information is specified using the relationship information, so that even if a situation such as a broken line of the character occurs, it is high. It becomes possible to judge the degree of blur with accuracy.

The first embodiment has been described above.
<2. Second Embodiment>
Next, a second embodiment will be described.

[2-1. system]
An example of a system to which the technology according to the second embodiment can be applied will be described. The technology according to the second embodiment can be applied to a system in which an object is photographed by an imaging device such as a digital camera and characters and symbols written on the object are automatically recognized using an image recognition technique.

(Character recognition system)
As an example, a character recognition system that recognizes characters from captured images will be described. FIG. 2 shows an example of a character recognition system that recognizes characters from an imaging device installed on the ground. In the following, the description will be made assuming the example of FIG. FIG. 2 is a diagram illustrating an example of a character recognition system according to the second embodiment. The character recognition system 100 shown in FIG. 2 is an example of a character recognition system according to the second embodiment.

  As shown in FIG. 2, the character recognition system 100 includes an imaging device 101 and an image processing device 102. The imaging device 101 is installed at a high position from the ground, for example, and captures characters written on an object such as a sign located in the vicinity. In the example of FIG. 2, the imaging range by the imaging device 101 is a range of the angle θ expressed by two broken lines. In addition, the imaging apparatus 101 is focused to focus on the imaging range. However, the depth of field is the range F, and even within the imaging range, the image of the object located within the range F is clear, but the image of the object located outside the range F is blurred.

For example, captured image P _A of the imaged object at a range F outside the point A becomes larger blurred image as shown in FIG. On the other hand, the captured image P _B of the object imaged at point B in the range F is than imaged image P _A captured at the point A becomes clear image. Even when an object within the range F is photographed, there is a possibility that the captured image is blurred due to subject blurring or temperature change. Further, the focus position may be shifted due to the deterioration of the imaging device 101 over time, and the captured image may be blurred.

When blur occurs in the captured image, it is difficult to recognize characters from the captured image of the object. Therefore, the image processing apparatus 102 detects the degree of blur of the captured image (hereinafter, input image P _IN ) input from the imaging apparatus 101, and executes character recognition processing according to the detected degree of blur. Hereinafter, the contents of the processing executed by the image processing apparatus 102 will be further described with reference to FIGS. 3 and 4. FIG. 3 is a first diagram for explaining the character recognition system according to the second embodiment. FIG. 4 is a second diagram for explaining the character recognition system according to the second embodiment.

Examples 3 and 4 relates to the case where the captured image P _A is input from the imaging device 101 to the image processing apparatus 102. In other words, this example relates to the case where the input image P _IN is captured image P _A. In the following, the magnitude of blur of an image is expressed as blur levels Lv1, Lv2, and Lv3. Further, Lv1 <Lv2 <Lv3. Here, for convenience of explanation, three types of blur conditions are illustrated, but the technique described below can also be applied to two types or four or more types of blur conditions.

The image processing apparatus 102 holds a plurality of dictionary images respectively corresponding to a plurality of blur levels Lv1, Lv2, and Lv3. The dictionary image is an image of characters prepared in advance for comparison with the character portion of the input image _PIN during character recognition. For example, if the blurriness of the input image P _IN is Lv3, the image processing apparatus 102 recognizes the characters included in the input image P _IN by comparing the input image P _IN and a dictionary image blurriness Lv3.

If blurriness of the input image P _IN is Lv3, as shown in FIG. 3, the similarity S _A of the image (A) of the input image P _IN blur degree Lv1 input image P _IN blur images degree Lv2 ( It is smaller than the similarity S _B and B). Furthermore, the similarity S _B of the image (B) of the input image P _IN blur degree Lv2 is smaller than the similarity S _C of the image (C) of the input image P _IN blur degree Lv3. That is, the accuracy of character recognition is improved by comparing the dictionary image with the blur level Lv3 and the input image _PIN to recognize characters included in the input image _PIN .

However, if each of the plurality of dictionary images is compared with the input image _PIN , the load on the comparison processing increases, so that the image processing apparatus 102 detects from the input image _{PIN as} shown in FIG. Select a dictionary image that matches the degree of blur, and collate both images.

As described above, the accuracy of character recognition is improved by using a dictionary image corresponding to the degree of blur of the input image _PIN . Furthermore, by selecting one dictionary image according to the determination result of the degree of blur, the processing load for calculating the similarity by comparing the input image _PIN with the dictionary image is reduced.

(Other systems)
The technology according to the second embodiment can be applied to, for example, a landmark recognition system or an OCR (Optical Character Recognition) system in addition to the character recognition system 100 described above.

  The landmark recognition system is a system that captures a signboard or the like in a town or a sightseeing spot with a camera function of a smartphone and provides information related to landmarks such as areas and facilities based on a character string recognized from the captured image. When the technology according to the second embodiment is applied to the landmark recognition system, even if a captured image is blurred due to a shooting situation or the like, character recognition is performed with high accuracy, and the possibility that correct information can be provided increases.

  The OCR system is a system that recognizes individual characters from an image of a character string read by an optical device such as a scanner and converts the characters into text data. When the technology according to the second embodiment is applied to the OCR system, the character of the read original is blurred and unclear, or the light transmitting plate (glass surface, transparent plastic surface, etc.) disposed opposite to the original surface is stained. Even in some cases, character recognition can be performed with high accuracy.

  Heretofore, an example of a system to which the technology according to the second embodiment can be applied has been described. In the following, description will be given by taking the character recognition system 100 as an example, but the scope of application of the technology according to the second embodiment is not limited to this. The technique according to the second embodiment can be applied to any system that performs character recognition from an image.

[2-2. Determining the degree of blur]
Described below is how to specify the degree of blur.
(Blur condition and blur index)
The image processing apparatus 102 recognizes characters by performing dictionary selection as described above. In this case, the degree of blur is specified. Hereinafter, the blur condition specifying operation by the image processing apparatus 102 will be described. First, the degree of blur and the blur index will be described.

FIG. 5 is a diagram for explaining the relationship between the degree of blur, the edge gradient, and the edge width. For example, an edge gradient, an edge width, or the like is used as a blur index serving as a scale representing the degree of blur. As shown in FIG. 5, when the luminance of the edge portion is graphed, an image (A) with a large degree of blur has a small edge gradient (angle η ₁ ) and an edge width (distance W ₁ from the minimum luminance to the maximum luminance). Is big. On the other hand, the image (B) with a small degree of blur has a large edge gradient (angle η ₂ ) and a small edge width (distance W ₂ from the minimum luminance to the maximum luminance). By using such characteristics, the degree of blur can be evaluated based on the edge gradient and the edge width.

  However, if the shooting area is large or the resolution of the characters included in the input image is low, the details of the input image will be lost and the details will be crushed. Sometimes.

  FIG. 6 is a diagram illustrating the luminance of the edge portion. (A) has shown the graph g1 of the brightness | luminance of the edge part in the image p1. The image p1 is a captured image of the character line L1 having a sufficient thickness so as not to be crushed by the image blur.

(B) shows a graph g2 of the luminance of the edge portion in the image p2. The image p2 is a captured image of the character line L2 that is thin enough to be crushed by the image blur.
The minimum luminance at the edge portion of the image p1 in (A) is b1. On the other hand, the minimum luminance at the edge portion of the image p2 in (B) is b2 (> b1), and there is a difference Δ in luminance between the images p1 and p2.

  As described above, when the character line is thin due to a wide-angle shooting area or a low resolution, the line may be crushed due to the blur of the input image, and a difference Δ may occur in the luminance of the original line. is there. As a result, correct edge gradient and edge width values as an index of the degree of blur cannot be obtained.

(Reason why the correct edge width cannot be calculated from the broken image)
FIG. 7 is a diagram showing the edge width. The graph g2 of the brightness | luminance of the edge part in the image p2 of FIG. 6 where the line collapse occurred is shown. The luminance at the edge portion of the image p2 is b2, and the edge width is W11. The luminance b1 is the original luminance of the character.

  If the edge width is calculated from the image p2, the edge width = W11 is calculated as shown in FIG. 7, but since the image p2 is an image in which the line is crushed, there is no line crushed. The minimum luminance is likely to be smaller than b2. In that case, the edge width is also different from W11. Therefore, it is desirable that the edge width calculated from the image p2 in which the line is crushed is not a correct value and not used as a criterion for determining the degree of blur.

  The edge width obtained from an image with a broken line has been described so far, but the application of an edge gradient obtained from an image with a broken line as an index of the degree of blur will be described. First, normalization for correcting the brightness of an image will be described.

  FIG. 8 is a diagram for explaining normalization. (A) shows the graph ga of the luminance of the edge portion in the image p3 and its normalized graph ga1. (B) shows a graph gb of the luminance of the edge portion in the image p4 and its normalized graph gb1.

Note that the edge widths of the images p3 and p4 are the same, but the brightness of the images is different from each other, and the image p3 is brighter than the image p4.
In the graph ga of (A), the edge width at the edge portion of the image p3 is W12, and the edge gradient is SL1. On the other hand, in the graph gb in (B), the edge width at the edge portion of the image p4 is W12, and the edge gradient is SL2.

  Thus, even if the edge widths of the images p3 and p4 are the same, if the brightness of the images is different, the edge gradient is different. In this example, the edge gradient SL1 of the image p3 is the edge gradient of the image p4. It is larger than SL2.

  As described above, when the edge widths that can be observed are the same, but the image brightness is different, normalization that corrects the different character luminance and background luminance to the reference character luminance and the reference background luminance, respectively. Is done.

  In both the graph ga1 in (A) and the graph gb1 in (B), the reference character luminance is corrected to br1 and the reference background luminance is corrected to br2. By this correction, the edge gradients of each other are equal to SL0. By performing such normalization, as shown in the graphs ga1 and gb1, the edge gradient can be set to the same value for images having the same edge width even if the brightness is different.

  FIG. 9 is a diagram for explaining normalization for a line-broken image. (A) shows a graph g11 of the luminance of the edge portion in the image p5 without line collapse. The luminance at the edge portion of the image p5 in (A) is b11, which is the original character luminance (that is, the same character luminance as when no crushing occurs).

  (B) shows a graph g12 of the luminance of the edge portion in the image p6 where the line is crushed. The luminance at the edge portion of the image p6 in (B) is b11a, which is the observed character luminance (that is, the character luminance when crushing occurs). The images p5 and p6 are images having the same brightness.

  When normalization is performed on an image with a broken line as in the image p6 in (B), normalization is performed based on observable character luminance instead of original character luminance. In this case, the original character luminance is not recognized, normalization fails, and the correct edge gradient cannot be calculated from the image p6 in which the line is crushed. For these reasons, it may not be possible to correctly calculate blur indices such as edge gradients and edge widths for images in which line collapse occurs and the line width is not correctly recognized. Therefore, it is desired to obtain an accurate line width.

(Circle of confusion)
Here, the circle of confusion representing the degree of blur will be described. A circle in which light emitted from a point light source spreads on the image plane is called a circle of confusion. As the subject moves away from the in-focus position, the circle of confusion increases.

  In addition, at the boundary portion between the character to be recognized and the background, the reflected light from the character portion and the reflected light from the background portion are incident on the pixels around the edge where the brightness of the boundary portion changes. . For this reason, the pixels in the circle of confusion at the boundary between the character and the background are pixels having intermediate brightness between light and dark, so the size of the circle of confusion (for example, the diameter of the circle of confusion) is the degree of blur. Represent.

(Relationship between edge gradient and circle of confusion)
Hereinafter, an example in which an edge gradient is used as a blur index is shown, but an edge width can also be used as a blur index. FIG. 10 is a diagram showing the relationship between the edge gradient and the circle of confusion. The vertical axis is the edge gradient, and the horizontal axis is the circle of confusion. Between the edge gradient and the circle of confusion, there is a relationship in which the circle of confusion increases as the edge gradient decreases. This relationship can be expressed in the form of a function (hereinafter referred to as a blur function). Of course, the above relationship can also be expressed using a table that associates the edge gradient with the size of the circle of confusion.

  The blur function or table can be obtained in the form of a function or a table by calculating the edge gradient and the size of the circle of confusion by experiment or simulation in advance and using the analytical method or the statistical method from the calculation result. For example, a method of specifying a function by a least square method, multivariate analysis, or the like, a method of holding the result of the calculation as a table, or the like can be applied.

If the above blur condition function is used, the size of the circle of confusion of the subject can be specified from the edge gradient calculated for the subject having a certain step edge.
However, when the line is thin, the line is crushed. Therefore, when the blur condition function calculated on the assumption of the step edge is applied, the relationship between the accurate edge gradient and the circle of confusion may not be established. In this case, there is a possibility that the size of the circle of confusion cannot be correctly specified from the edge gradient.

  Therefore, in the second embodiment, in addition to the variable of the size of the circle of confusion, the line width is also added as a variable, and the blur condition function for each line width representing the relationship between the circle of confusion, the line width, and the edge gradient is applied. . Of course, the relationship can also be expressed in a table format.

  The above blur condition function or table calculates the relationship between the size of the circle of confusion and the edge gradient in advance for each line width by experiment or simulation, and uses the calculation result in the form of a function or table by an analytical method or a statistical method. Can be sought. For example, a method of specifying a function by a least square method, multivariate analysis, or the like, a method of holding the result of the calculation as a table, or the like can be applied.

  FIG. 11 is a diagram showing the relationship between the line width, the edge gradient, and the circle of confusion. The blur condition function f1 represents the correspondence between the edge gradient in the line width W1a and the circle of confusion, and the blur condition function f2 represents the correspondence between the edge gradient in the line width W2a and the circle of confusion. The blur condition function f3 represents the correspondence between the edge gradient in the line width W3a and the circle of confusion. Of the line widths W1a, W2a, and W3a, the line width W1a is the smallest and the line width W3a is the largest (W1a <W2a <W3a).

  In this way, for each different line width, the relationship between the edge gradient and the size of the circle of confusion is obtained and stored in a database. And when identifying the circle of confusion from the input image, if you know the line width when the characters in the input image are not blurred, you can know which correspondence in the database should be used, so the confusion from the edge gradient The size of the circle can be specified.

(About calculation of line width)
A method for calculating the line width in the unblurred state from the blurred line in the input image will be described below. It is assumed that a letter L is written in a certain area R. In this case, the ratio of the size of the region R to the line width of the character L is determined, and when the size is measurable and sufficiently larger than the line width, the character L written in the region R It is possible to calculate the line width for.

  FIG. 12 is a diagram illustrating an example of a region used for calculating the line width. When a character is written in a certain area and the width of the character line is determined with respect to the size of the area, it is written from the width of the captured image P11 of such an area (hereinafter also referred to as a sign). The line width of a character can be specified.

  In this way, there is an object or target area in which the ratio of the size to the line width is determined, and the size is measured on the precondition that the size can be measured and is sufficiently larger than the line width. Thus, the line width can be calculated relatively.

FIG. 13 is a diagram for explaining the calculation of the line width. (A) shows the calculated value X1 of the line width of the character observed directly from the image, and (B) shows the calculated value X2 of the sign size.
In (A), when trying to recognize a character on a sign with a fixed style, the thickness of the line when the character appears unblurred is A1 [pixel], and the change in line width due to blur is B1 [ When the pixel width observation error due to noise is C1 [pixel], the line width measurement value X1 observed directly from the image is calculated by the following equation (1).

X1 = A1 + B1 + C1 (1)
In (B), when the sign size (width) is A2 [pixel], the measurement error of the sign size due to blur is B2 [pixel], and the measurement error of the sign size due to noise is C2 [pixel], The measured value X2 of the marker size is calculated by the following equation (2).

X2 = A2 + B2 + C2 (2)
It should be noted that the amounts of blur are substantially equal since they are in the same plane, and the observation errors due to noise can be regarded as substantially equal, so B2≈B1 and C2≈C1. Further, since the ratio between the width of the sign and the line width of the character is determined, if the proportionality constant is M, the relationship between A1 and A2 is expressed by the following equation (3).

A2 = M × A1 (3)
Here, a method for calculating the line width A1 in a non-blurred state will be described. Since A1 = X1− (B1 + C1) from the above formula (1), when X1 is observed and used as the calculated value of the line width A1, B1 + C1 becomes an error.

Further, the following formula (4) is obtained from the above formulas (2) and (3).
A1 = A2 / M = (X2 / M)-(B2 + C2) / M (4)
Then, from B2≈B1 and C2≈C1, Expression (4a) is obtained.

A1≈ (X2 / M) − (B1 + C1) / M (4a)
When X2 is observed and used as the calculated value of the line width A1, (B1 + C1) / M becomes an error as shown in the equation (4a).

  Here, (B1 + C1) is often a few pixels at most. Therefore, the value of the proportionality constant M is larger than the value of (B1 + C1), and (B1 + C1) / M << B1 + C1. Therefore, when calculating the line width, the error is smaller when X2 is measured and converted by the proportional constant M than when X1 is measured, and the original line width is calculated more accurately. Can do.

The identification of the degree of blur has been described above.
[2-3. Function of image processing apparatus]
Next, the function of the image processing apparatus 102 will be further described with reference to FIGS.

  FIG. 14 is a block diagram illustrating an example of the functions of the image processing apparatus according to the second embodiment. FIG. 15 is a diagram showing an example of confusion circle specifying information according to the second embodiment. FIG. 16 is a diagram illustrating an example of dictionary information according to the second embodiment.

  As shown in FIG. 14, the image processing apparatus 102 includes an image acquisition unit 111, a recognition target detection unit 112, a blur index calculation unit 113, a storage unit 114, a confusion circle specifying information selection unit 115, a confusion circle specifying unit 116, and a dictionary selection. Unit 117 and character recognition unit 118.

  The image acquisition unit 111 acquires a monitoring range image corresponding to a range preset as a monitoring target from the imaging apparatus. The recognition target detection unit 112 detects a sign from the image using an image processing technique, and cuts out a region (hereinafter referred to as a sign image) based on the detected sign. For example, a technique described in Japanese Patent Application Laid-Open No. 2008-276291 can be applied to the cutout. The blur index calculation unit 113 calculates a blur index from the clipped marker image. For example, the edge width or edge gradient at the boundary between the character portion and the background portion can be used as a blur index.

  The storage unit 114 stores confusion circle specifying information 114a and dictionary information 114b. The circle of confusion specific information 114a stores information indicating the relationship (blurring condition function) between the circle of confusion and the blur index corresponding to each line width described with reference to FIG. The information may be stored as function format information (that is, a combination of a function format and a coefficient), or may be stored as table format information. In the latter case, for example, conversion tables T1-1 to T1-3 as shown in FIG. 15 are prepared. The conversion tables T1-1 to T1-3 are table information indicating the correspondence between the edge gradients in the line widths W1a, W2a, and W3a and the size of the circle of confusion.

Further, the dictionary information 114b is information associating characters and dictionary images as shown in FIG. A dictionary image is prepared for each of a plurality of preset blur conditions. In the example of FIG. 16, as the dictionary image corresponding to the character “4”, the dictionary image P _D1 corresponding to the blur condition Lv1, the dictionary image P _D2 corresponding to the blur condition Lv2, and the dictionary image P _D3 corresponding to the blur condition Lv3 are obtained. It is prepared. A set of dictionary images prepared for each degree of blur is called a dictionary. In the example of FIG. 16, the dictionary D ₁ is a set of dictionary images P _D1 corresponding to the characters “4”, “6”, “9” _,.

Refer to FIG. 14 again.
The circle of confusion specific information selection unit 115 calculates line width and refers to the circle of confusion specific information 114a to determine which information (conversion table) to use when specifying the circle of confusion from the line width and the blur index. select.

  The confusion circle specifying unit 116 uses the conversion table selected by the confusion circle specifying information selecting unit 115 to specify the circle of confusion (the degree of blur of imaging) from the blur index obtained by the blur index calculating unit 113. For example, in FIG. 15, when the line width calculated by the confusion circle specifying information selection unit 115 is the line width W3a and the edge gradient calculated by the blur index calculation unit 113 is 60, the conversion table T1-3 causes the confusion. The size of the circle is specified as 4.

  The dictionary selection unit 117 selects a dictionary suitable for character recognition from the circle of confusion specified by the confusion circle specifying unit 116 and notifies the character recognition unit 118 of the selected dictionary. For example, if the circle of confusion is two pixels, a dictionary in which two pixels are blurred with respect to a dictionary in which the character portion and the background portion are divided into two values is selected.

  The above dictionary may be selected by preparing a blurred one, or only a dictionary that is not blurred may be prepared and generated each time. In the former case, processing time is not required, but the memory size to be used increases. In the latter case, it is only necessary to store a non-blurred dictionary in the memory, so that the memory size can be reduced. On the other hand, the processing takes time for executing the processing for generating the dictionary. The blurred dictionary can be generated, for example, by applying a Gaussian filter to a dictionary that is not blurred.

The character recognition unit 118 calculates the similarity between the dictionary selected by the dictionary selection unit 117 and the character of the input image, and outputs the dictionary with the highest similarity as the recognition result.
(Operation of the circle of confusion specific information selection unit 115)
Here, the operation of the confusion circle specifying information selection unit 115 will be described.

FIG. 17 is a flowchart showing the operation of the circle of confusion specific information selection unit.
[S1] The circle of confusion specific information selection unit 115 measures the width (number of pixels) of the entire sign from the position information of the sign detected by the recognition target detection unit 112.

  [S2] The circle of confusion specific information selection unit 115 calculates the line width of the character on the sign when it is not blurred. For example, it is assumed that the sign form to be recognized is defined as the sign width R2 [mm] and the character line width R1 [mm] (R2> R1), and the measured sign width is X2 [ pixel], the line width X1 [pixel] in a state where the characters on the sign are not blurred is calculated as (R1 / R2) · X2.

  [S3] The confusion circle specifying information selection unit 115 selects a conversion table that is the confusion circle specifying information 114a. In the confusion circle specifying information 114a in the storage unit 114, the relationship between the confusion circle and the blur index corresponding to each line width (for example, table information shown in FIG. 15) is recorded in advance. Since the line width X1 in the unblurred state is calculated in the process of S2, a conversion table of the relationship between the circle of confusion corresponding to the line width X1 and the blur index is used.

The function of the image processing apparatus 102 has been described above.
[2-4. Image processing device hardware]
Here, an example of hardware capable of realizing the function of the image processing apparatus 102 will be described with reference to FIG. FIG. 18 is a diagram illustrating an example of hardware capable of realizing the functions of the image processing apparatus according to the second embodiment.

  The functions of the image processing apparatus 102 can be realized by using, for example, hardware resources of the information processing apparatus illustrated in FIG. That is, the functions of the image processing apparatus 102 are realized by controlling the hardware shown in FIG. 18 using a computer program. In addition, depending on the embodiment, a modification in which a part of the hardware illustrated in FIG. 18 is omitted or another element is newly added is allowed.

  As shown in FIG. 18, this hardware mainly includes a CPU 902, a ROM (Read Only Memory) 904, a RAM 906, a host bus 908, and a bridge 910. Further, this hardware includes an external bus 912, an interface 914, an input unit 916, an output unit 918, a storage unit 920, a drive 922, a connection port 924, and a communication unit 926.

  The CPU 902 functions as, for example, an arithmetic processing unit or a control unit, and controls the overall operation of each component or a part thereof based on various programs recorded in the ROM 904, the RAM 906, the storage unit 920, or the removable recording medium 928. . The ROM 904 is an example of a storage device that stores a program read by the CPU 902, data used for calculation, and the like. The RAM 906 temporarily or permanently stores, for example, a program read by the CPU 902 and various parameters that change when the program is executed.

  These elements are connected to each other via, for example, a host bus 908 capable of high-speed data transmission. On the other hand, the host bus 908 is connected to an external bus 912 having a relatively low data transmission speed via a bridge 910, for example. As the input unit 916, for example, a mouse, a keyboard, a touch panel, a touch pad, a button, a switch, a lever, or the like is used. Furthermore, as the input unit 916, a remote controller capable of transmitting a control signal using infrared rays or other radio waves may be used.

  As the output unit 918, for example, a display device such as a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display), a PDP (Plasma Display Panel), or an ELD (Electro-Luminescence Display) is used. As the output unit 918, an audio output device such as a speaker or headphones, or a printer may be used. In other words, the output unit 918 is a device that can output information visually or audibly.

  The storage unit 920 is a device for storing various data. As the storage unit 920, for example, a magnetic storage device such as an HDD is used. Further, as the storage unit 920, a semiconductor storage device such as an SSD (Solid State Drive) or a RAM disk, an optical storage device, a magneto-optical storage device, or the like may be used.

  The drive 922 is a device that reads information recorded on a removable recording medium 928 that is a removable recording medium or writes information on the removable recording medium 928. As the removable recording medium 928, for example, a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is used.

  The connection port 924 is a port for connecting an external connection device 930 such as a USB (Universal Serial Bus) port, an IEEE 1394 port, a SCSI (Small Computer System Interface), an RS-232C port, or an optical audio terminal. For example, a digital camera or a printer is used as the external connection device 930.

  The communication unit 926 is a communication device for connecting to the network 932. As the communication unit 926, for example, a communication circuit for wired or wireless LAN (Local Area Network), a communication circuit for WUSB (Wireless USB), a communication circuit or router for optical communication, an ADSL (Asymmetric Digital Subscriber Line) Communication circuits, routers, communication circuits for mobile phone networks, and the like are used. A network 932 connected to the communication unit 926 is a wired or wireless network, and includes, for example, the Internet, a LAN, a broadcast network, a satellite communication line, and the like.

  With the hardware configuration as described above, the processing functions of this embodiment can be realized. Further, when the processing functions of the present embodiment are realized by a computer, a program describing the processing contents of the functions of the image processing apparatus 102 is provided.

  By executing the program on a computer, the above processing functions are realized on the computer. The program describing the processing contents can be recorded on a computer-readable recording medium. Examples of the computer-readable recording medium include a magnetic storage device, an optical disk, a magneto-optical recording medium, and a semiconductor memory. Examples of the magnetic storage device include a hard disk device (HDD), a flexible disk (FD), and a magnetic tape. Optical discs include DVD, DVD-RAM, CD-ROM / RW, and the like. Magneto-optical recording media include MO (Magneto Optical disk). The recording medium for recording the program does not include a temporary propagation signal itself.

  When distributing the program, for example, a portable recording medium such as a DVD or a CD-ROM in which the program is recorded is sold. It is also possible to store the program in a storage device of a server computer and transfer the program from the server computer to another computer via a network.

  The computer that executes the program stores, for example, the program recorded on the portable recording medium or the program transferred from the server computer in its own storage device. Then, the computer reads the program from its own storage device and executes processing according to the program. The computer can also read the program directly from the portable recording medium and execute processing according to the program. Further, each time the program is transferred from the server computer, the computer can sequentially execute processing according to the received program.

[2-5. Process flow]
Here, the operation of the image processing apparatus 102 will be further described with reference to FIG. FIG. 19 is a flowchart showing the operation of the image processing apparatus according to the second embodiment. For convenience of explanation, it is assumed that conversion tables T1-1, T1-2, and T1-3 (see FIG. 15) are stored in the storage unit 114 as the confusion circle specifying information 114a.

  [S101] The image acquisition unit 111 acquires a monitoring range image corresponding to a range set in advance as a monitoring target from the imaging apparatus. The recognition target detection unit 112 detects a sign from the image using an image processing technique, and cuts out a region (a sign image) based on the detected sign.

  [S102] The blur index calculation unit 113 calculates a blur index from the sign image cut out in S101. As the blur index, for example, the edge width or edge gradient of the boundary between the character portion and the background portion can be applied.

[S103] The circle of confusion specific information selection unit 115 calculates the line width of the character.
For example, the measured value of the size of the sign image is X2 [pixel], the thickness of the line when the character is not blurred is A1 [pixel], the change in the line width due to blur is B1 [pixel], and the noise The observation error of character width is expressed as C1 [pixel]. Also, the measured line width measured directly from the sign image is X1, the sign image size is A2 [pixel], the sign image size measurement error is B2 [pixel], and the sign image size measurement error is Indicated as C2 [pixel]. In this case, the circle of confusion specific information selection unit 115 can obtain A1 as the line width by the above equation (4a).

  [S104] The circle of confusion specific information selection unit 115 determines whether the line width calculated in S103 is W1a, W2a, or W3a. If the line width is W1a, the process proceeds to S105. If the line width is W2a, the process proceeds to S106. If the line width is W3a, the process proceeds to S107. When a conversion table corresponding to another line width is prepared as the confusion circle specifying information 114a, the determination process is similarly performed for the line width corresponding to the conversion table.

  [S105] The circle of confusion specific information selection unit 115 selects the conversion table T1-1. The confusion circle specifying unit 116 uses the conversion table selected by the confusion circle specifying information selecting unit 115 to specify the size of the confusion circle corresponding to the blur index calculated in S102. When the process of S105 is completed, the process proceeds to S108.

  [S106] The circle of confusion specific information selection unit 115 selects the conversion table T1-2. The confusion circle specifying unit 116 uses the conversion table selected by the confusion circle specifying information selecting unit 115 to specify the size of the confusion circle corresponding to the blur index calculated in S102. When the process of S106 is completed, the process proceeds to S108.

  [S107] The circle of confusion specific information selection unit 115 selects the conversion table T1-3. The confusion circle specifying unit 116 uses the conversion table selected by the confusion circle specifying information selecting unit 115 to specify the size of the confusion circle corresponding to the blur index calculated in S102. When the process of S107 is completed, the process proceeds to S108.

  [S108] The dictionary selection unit 117 refers to the dictionary information 114b (see FIG. 16) and selects a dictionary corresponding to the size of the circle of confusion identified in S105, S106, or S107. For example, if the circle of confusion is two pixels, the dictionary selection unit 117 selects a dictionary in which two pixels are blurred with respect to a dictionary in which a character part and a background part are divided into two values.

  [S109] The character recognition unit 118 calculates the similarity between the dictionary selected in S108 and the character included in the sign image, and outputs the character corresponding to the dictionary image with the highest similarity as the recognition result. When the process of S109 is completed, the series of processes shown in FIG.

The processing flow relating to the operation of the image processing apparatus 102 has been described above.
As described above, according to the present technology, the accuracy of specifying the degree of blur is improved, and as a result, it is possible to improve the accuracy of the image processing method using the result of specifying the degree of blur. Further, when the dictionary is selected according to the degree of blur, the recognition accuracy is improved, and when the focus is adjusted so that the degree of blur is reduced, the focus adjustment accuracy can be improved.

  As mentioned above, although embodiment was illustrated, the structure of each part shown by embodiment can be substituted by the other thing which has the same function. Moreover, other arbitrary structures and processes may be added.

DESCRIPTION OF SYMBOLS 10 Information processing apparatus 11 Memory | storage part 12 Calculation part F1, F2, F3 Function curve Wa, Wb, Wc Line width y Blur parameter | index x Blur condition

Claims (6)

A storage unit that stores relationship information indicating a relationship between a blur index that is an index of a blur condition of the image and the blur condition according to a line width of a character included in the image;
A calculation unit that calculates the line width and the blur index from a captured image, refers to the relation information stored in the storage unit, and specifies the blur condition corresponding to the line width and the blur index;
An information processing apparatus. The arithmetic unit measures the size of an object in the image in which a ratio between the line width and the size is determined, and calculates the line width in a state where the ratio is not blurred from the measured ratio. The information processing apparatus described in 1. The information processing apparatus according to claim 1, wherein the calculation unit detects an edge portion of the character and uses a gradient of the edge portion or a width of the edge portion as the blur index. The storage unit stores a plurality of dictionary images that are images used for recognition of the characters,
The information processing apparatus according to claim 1, wherein the calculation unit selects the dictionary image according to the specified degree of blur, and recognizes the character using the selected dictionary image. Computer
According to the line width of the character included in the image, the relationship information is acquired from a storage unit that stores relationship information indicating a relationship between a blur index that is a blur condition index of the image and the blur condition,
The line width and the blur index are calculated from the captured image, and the blur condition corresponding to the line width and the blur index is specified with reference to the relation information stored in the storage unit. Method. On the computer,
According to the line width of the character included in the image, the relationship information is acquired from a storage unit that stores relationship information indicating a relationship between a blur index that is a blur condition index of the image and the blur condition,
The line width and the blur index are calculated from the captured image, and the blur information corresponding to the line width and the blur index is specified with reference to the relation information stored in the storage unit. Let the program.

Images