JPH08194820A - Picture processor and picture processing method - Google Patents

Abstract

PURPOSE: To provide a picture processor and picture processing method capable of efficiently extracting a picture element having a density gradation vector with desired azimuth. CONSTITUTION: A detection range between 1st and 2nd azimuth parts is specified by a specifying means 1. An auxiliary vector setting means 2 sets two auxiliary vectors respectively rectangular to the 1st and 2nd azimuth parts with (x) and (y) components. A density gradation calculating means 3 calculates the density gradation vector of each picture element as respective component differential values Gx, Gy in the (x) and (y) axes based upon respective density values of each picture element and its peripheral picture elements. An inner product means 4 calculates respective inner products of the density gradation vector and two auxiliary vectors in each picture element. A judging means 5 judges whether the azimuth of the density gradation vector of each picture element is included in the detection range or not base upon the codes of respective inner product values.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to improvements in an image processing apparatus and an image processing method for processing an image, and in particular,
The present invention relates to efficiently detecting pixels having a density gradient vector in a desired range.

[0002]

2. Description of the Related Art In recent years, with the progress of computers, image processing apparatuses and image processing methods for processing various images have been remarkably spread. Although various types of images to be subjected to image processing are known, a density gradient vector can be calculated in all parts of an image having shading. The density gradient vector is one of the image feature quantities that represent the direction and magnitude of the grayscale change in each pixel. The density gradient vector of a pixel is determined based on the gray value of pixels around the pixel and is useful for catching a local change in the image gray value.

If a pixel having a density gradient vector of an arbitrary azimuth can be extracted, it can be applied to the processing of a texture image. For example, if a material having a pattern in a certain direction is photographed by a camera and a pixel having a density gradient vector in a direction different from the original pattern is detected, a flaw such as a crack can be found. Such applications contribute to the sophistication of control of FA (factory automation) and the like. In addition, the extraction of pixels having an arbitrary density gradient vector can be widely applied to medical images and OCR.

Here, a typical image density gradient calculation and a method of detecting the azimuth angle, that is, the direction of the density gradient vector in the conventional image processing apparatus and image processing method will be described. First, the density gradient at a predetermined pixel coordinate (i, j) in the input image is first obtained from the differential values in the X and Y directions calculated from the gray values of several surrounding pixels centering on that pixel. To be

Here, FIG. 7 is a principle diagram showing a specific example of calculation of the concentration gradient vector. That is, each of the X component and the Y component of the density gradient of the pixel coordinate (i, j) is G
A vector having x and Gy and these components is called a density gradient vector of this pixel. In FIG. 7, X1 to X
Up to 9 represent the image gray value of the pixel at the pixel coordinate (i, j) and the adjacent 8 pixels centered on the pixel. At this time, Gx and Gy are, for example, as shown in the figure,

## EQU1 ## Gx = (X3 + X6 + X9)-(X1 + X4 + X7) Gy = (X1 + X2 + X3)-(X7 + X8 + X9). The magnitude of the concentration gradient vector at this time is

[Equation 2] The azimuth angle of the concentration gradient vector is

## EQU3 ## It is represented by tan ^-1 (Gy / Gx). However, tan-1 (Gy / Gx) is in the range of 0 [radian] to 2π [radian].

In this case, in order to detect a pixel having a density gradient vector having an azimuth angle within a predetermined range, the conventional technique has generally performed as follows. That is, first, the detection range is given in angle (radian). The azimuth angles of the density gradient vectors for all pixels are calculated in advance in radians, which is the same format as the detection range. As described above, conventionally, both comparison targets are expressed in radians of the same format, and then the angle comparison calculation is performed to extract pixels that meet the conditions.

As another method, the inner product of the density gradient vector obtained for each pixel of the image and one or two fixed reference vectors is calculated, and the obtained inner product value is set to the magnitude. In contrast, there is a method in which weighting is performed according to the gray value of each pixel and only pixels having a certain value or more are extracted (Japanese Patent Laid-Open No. 63-81579).

[0008]

However, the above-mentioned conventional techniques have the following problems.
That is, in the method of calculating the azimuth angle of the density gradient vector in each pixel in advance, it is necessary to calculate the inverse trigonometric function for all pixels in order to obtain the azimuth angle. Since this calculation is performed by a floating point calculation, it takes time, which makes it difficult to detect it efficiently. Further, when comparing the magnitudes of angles, a large number of conditional branches are required due to the periodicity of 2π [radian], which makes it difficult to improve the detection efficiency and complicates the configuration.

On the other hand, in the method based on the inner product of the fixed reference vector and the density gradient vector, the density gradient vector in any direction cannot be searched because the reference vector is fixed.

The present invention has been proposed to solve the above-mentioned problems of the prior art, and its object is to perform image processing for efficiently extracting pixels having a density gradient vector of a desired azimuth angle. An apparatus and an image processing method are provided. Further, another object of the present invention is to provide an image processing apparatus and an image processing method in which processing such as confirmation and processing of a detection result is easy. Another object of the present invention is to provide an image processing apparatus and an image processing method that facilitate the processing that is suitable for the purpose.

[0011]

In order to achieve the above-mentioned object, the invention of claim 1 forms a density gradient of a desired azimuth angle from each pixel forming an image on the x, y plane and having a gray value. In an image processing device that detects pixels having a vector, a detection range that is a range of azimuth angles of a density gradient vector to be detected and that can be expressed as a range sandwiched between a first azimuth angle and a second azimuth angle is specified. Designating means, auxiliary vector setting means for setting two auxiliary vectors respectively orthogonal to the first azimuth angle and the second azimuth angle by an x component and ay component, and the pixel for each pixel And a density gradient calculating means for calculating the density gradient vector as the component differential values Gx, Gy of the x and y axes based on the gray values of the surrounding pixels, and the density gradient vector and the two complements for each pixel. An inner product means for calculating each inner product value with the vector, and a determination means for determining whether or not the azimuth angle of the concentration gradient vector of the pixel is included in the detection range based on the sign of each inner product value. It is characterized by having.

The invention according to claim 2 is the image processing apparatus according to claim 1, characterized in that it has parallel processing means for processing in parallel the calculation of two inner product values for each auxiliary vector.

According to a third aspect of the present invention, in the image processing apparatus according to the first aspect, the inner product means first calculates only the inner product value of the density gradient vector of the pixel and any one of the auxiliary vectors. It is determined that the azimuth angle of the concentration gradient vector of the pixel is included in the detection range side with respect to the first or second azimuth angle corresponding to the auxiliary vector based on the sign of the inner product value. When it is determined by the means, the inner product value with the other auxiliary vector is calculated.

According to a fourth aspect of the present invention, in the image processing apparatus according to the first aspect, each component value of each auxiliary vector is represented by an integer value in a predetermined range represented by a plurality of bits. It is characterized by that.

According to a fifth aspect of the invention, in the image processing apparatus according to the first aspect, there is provided a first storage means for sequentially storing the detected address value of the pixel on the memory.

According to a sixth aspect of the present invention, in the image processing apparatus according to the first aspect, there is provided a second storage means for sequentially storing the detected coordinate values on the image of the pixel.

According to a seventh aspect of the invention, in the image processing apparatus according to the first aspect, there is provided first conversion means for converting the address value of the pixel on the memory into the coordinate value of the pixel on the image. And

The invention according to claim 8 is the image processing apparatus according to claim 1, further comprising a second conversion means for converting the coordinate value on the image of the pixel into the address value on the memory of the pixel. And

According to a ninth aspect of the present invention, in the image processing apparatus according to the first aspect, there is provided a third storage means for storing the image of the detected pixel portion.

According to a tenth aspect of the present invention, in the image processing apparatus according to the first aspect, there is provided first weighting means for weighting each extracted pixel based on the magnitude of the gray value. To do.

Further, the invention of claim 11 is the image processing apparatus according to claim 1, which further comprises second weighting means for weighting each extracted pixel based on the magnitude of the density gradient vector. And

According to a twelfth aspect of the present invention, in the image processing apparatus according to the first aspect, for each extracted pixel,
It is characterized by further comprising third weighting means for weighting based on the difference between the two inner product values for each pixel.

According to the image processing method of claim 13, x,
In an image processing method for detecting a pixel having a density gradient vector of a desired azimuth angle from each pixel that forms an image on the y-plane and has a gray value, a range of azimuth angles of the density gradient vector to be detected, A step of designating a detection range that can be expressed as a range sandwiched between the first azimuth angle and the second azimuth angle, and two steps that are orthogonal to the first azimuth angle and the second azimuth angle. Auxiliary vector setting step of setting the auxiliary vector by x component and y component respectively, and for each pixel, the density gradient vector is differentiated for each component on the x and y axes based on each gray value of the pixel concerned and surrounding pixels. A step of calculating a density gradient for calculating the values Gx and Gy; a step of calculating an inner product of the density gradient vector and the two auxiliary vectors for each pixel; Wherein the azimuth angle of the gradient vector of the pixel based on the sign of the value comprises a step of determining determines whether or not included in the detection range.

The invention according to claim 14 is the image processing method according to claim 13, characterized in that it includes a step of parallel processing for parallel processing of calculation of two inner product values for each auxiliary vector. .

According to a fifteenth aspect of the present invention, in the image processing method according to the thirteenth aspect, in the step of the inner product,
For the density gradient vector of the pixel, only the inner product value with any one of the auxiliary vectors is first calculated, and the azimuth angle of the density gradient vector of the pixel corresponds to the auxiliary vector based on the sign of the inner product value. When it is determined in the determination step that it is included in the detection range side with respect to the first or second azimuth angle, the inner product value with the other auxiliary vector is calculated.

According to a sixteenth aspect of the present invention, in the image processing method according to the thirteenth aspect, each component value of each auxiliary vector is represented by an integer value in a predetermined range represented by a plurality of bits. To do.

The invention according to claim 17 is the image processing method according to claim 13, characterized in that it includes a first storing step for sequentially storing the address value on the memory of the detected pixel. .

The invention according to claim 18 is the image processing method according to claim 13, characterized in that it includes a second storage step of sequentially storing the coordinate values of the detected pixels on the image. .

The invention according to claim 19 is the image processing method according to claim 13, which further includes a first conversion step for converting the address value of the pixel on the memory into the coordinate value on the image of the pixel. Characterize.

The invention according to claim 20 is the image processing method according to claim 13, further comprising a second conversion step for converting the coordinate value of the pixel on the image into the address value of the pixel on the memory. Characterize.

Further, the invention of claim 21 is the image processing method according to claim 13, characterized by including a third storing step of storing an image of the detected pixel portion.

The invention according to claim 22 is the image processing method according to claim 13, characterized in that it includes a first weighting step for weighting each extracted pixel based on the magnitude of the grayscale value. And

The invention according to claim 23 is the image processing method according to claim 13, further comprising a second weighting step for weighting each extracted pixel based on the magnitude of the density gradient vector. Characterize.

The invention of claim 24 is the image processing method according to claim 13, wherein each extracted pixel is weighted based on the difference between the two inner product values of each pixel. It is characterized by including a step.

[0035]

The present invention having the above structure has the following functions. That is, in the inventions of claims 1 and 13, the detection range is a range having a certain width and can be expressed as a range sandwiched between the first azimuth angle and the second azimuth angle. Here, two auxiliary vectors are used. Each auxiliary vector is at right angles to the first azimuth angle and the second azimuth angle, respectively,
Each auxiliary vector is represented by an x component and ay component, respectively. On the other hand, for each pixel, the density gradient vector is calculated as the component differential values Gx and Gy of the x and y axes based on the gray values of the pixel and the surrounding pixels. The density gradient vector and the two auxiliary vectors are divided into two for each pixel.
Two dot product values are calculated.

Then, based on the sign of each inner product value, it is determined whether or not the density gradient vector of the pixel is included in the detection range. Here, in general, the inner product value between two vectors is positive when the angle between the azimuth angles of each vector is acute and negative when the angle is obtuse. Then, each auxiliary vector is orthogonal to the first azimuth angle and the second azimuth angle. Therefore, regarding the concentration gradient vector, the angular relationship between the first azimuth angle and the second azimuth angle can be specified by the sign of the inner product value with each auxiliary vector. For example, when the first auxiliary vector is on the second azimuth side with respect to the first azimuth and the second auxiliary vector is on the second azimuth side with respect to the first azimuth, the inner product When both the signs of the values are positive, the concentration gradient vector exists within the detection range.

The inner product value can be calculated by a simple operation of summing the products of x elements and the products of y elements, and the right angle can be determined based on only simple information such as a sign. As described above, in the inventions of claims 1 and 13, a simple information is detected without detecting a pixel having a density gradient vector of a desired azimuth angle by calculating the azimuth angle of the density gradient vector of all images in radians. And processing can be done efficiently.

Further, in the inventions of claims 2 and 14, since the calculation of the two inner product values for each auxiliary vector is performed in parallel, the detection processing is speeded up.

Further, in the invention of claims 3 and 15, only the inner product value with any one of the auxiliary vectors is first calculated,
Based on the sign of this inner product value, the azimuth angle of the density gradient vector of the pixel corresponds to the auxiliary vector.
When it is determined that the azimuth angle is included in the detection range side, the inner product value with the other auxiliary vector is calculated. Therefore, when it is determined based on the previously calculated inner product value that it is out of the detection range, the subsequent calculation of the inner product value can be omitted, and the number of calculations can be saved, thus speeding up the process.

In the inventions of claims 4 and 16, each component value of each auxiliary vector is represented by an integer value within a predetermined range. In general, the integer value has a simple internal representation format, and the arithmetic processing is easy and fast, so that the processing is speeded up.

Further, in the inventions of claims 5 and 17, since the address values of the detected pixels on the memory are sequentially stored and accumulated, the coordinate values of the image are set when accessing the gray value of the detected pixels. There is no need to perform the process of converting the address value again. Therefore, processing such as access to the image data after detection is speeded up.

Further, in the inventions of claims 6 and 18, since the coordinate values on the image of the detected pixels are sequentially stored, the processing using the coordinates such as imaging the detected portion and specifying the position is performed. Also in this case, it is not necessary to perform the process of converting the address value on the memory into the coordinate value on the image again.

Further, in the inventions of claims 7 and 19, even if the detected portion is stored in the form of the address value on the memory, since this address value is converted into the coordinate value on the image of the pixel, It is easy to perform processing using coordinates such as imaging the detected portion and specifying the position.

Further, in the inventions of claims 8 and 20, even if the detected portion is stored in the form of coordinate values on the image,
Since this coordinate value is converted into the address value on the memory,
The processing using the address value is easy when accessing the gray value of the detected portion.

Further, in the inventions of claims 9 and 21, since the detected pixel portion is stored as it is as the image type, it is necessary to perform the image forming process again when the image forming is necessary for confirmation. There is no.

Further, in the inventions of claims 10 and 22, each extracted pixel is weighted based on the magnitude of the gray value. For this reason, when the gray value is significant, such as when the dark portion of the image is not the object to be processed and the bright portion is important, it is easy to perform the processing that emphasizes the pixels according to the purpose of the processing.

According to the eleventh and twenty-third aspects of the present invention, each extracted pixel is weighted based on the magnitude of the density gradient vector. For this reason, when the magnitude of the density gradient vector is significant, such as when the density gradient is steep at the contour portion of the image, it becomes easy to perform processing that emphasizes pixels according to the purpose of the processing.

According to the twelfth and twenty-fourth aspects of the present invention, each extracted pixel is weighted based on the difference between two inner product values of the pixel. Here, in general, the smaller the difference between the inner product values, the closer the density gradient vector of the pixel and the central azimuth angle of the detection range. For this reason, it becomes easy to perform processing suitable for the purpose by giving priority to the processing of the pixel having the density gradient vector located in the center of the detection range.

[0049]

Embodiments of the present invention will now be specifically described with reference to the drawings. The embodiments described below are realized on a computer, and each function of the embodiments is realized by a predetermined procedure (software) controlling the computer.

Therefore, in the present specification, an embodiment will be described below by assuming a virtual circuit block (means) having each function of the embodiment. That is, each "means" or "unit" in this specification is a concept corresponding to each function of the embodiment, and does not necessarily correspond to a specific hardware or software routine on a one-to-one basis. The same hardware element may in some cases constitute different means. For example, a computer may be one means for executing one instruction and another for executing another instruction. Further, one means may be realized by only one instruction or may be realized by a large number of instructions. However, use of a computer is an example, and all or part of the functions of the present invention may be realized on an electronic circuit such as a custom chip (dedicated integrated circuit) if possible.

The computer used in the embodiment generally has a CPU (central processing unit) and a main memory device composed of a RAM (random writing / reading type memory element). The size of the computer is arbitrary, and any size such as a personal computer, a workstation, a mainframe, etc. may be used.

The hardware of the computer is typically an input device such as a keyboard and a mouse, an external storage device such as a hard disk device, and a CRT.
It includes an output device such as a display device and a printer printer, and necessary input / output control circuits.

However, the hardware configuration is arbitrary, and as long as the present invention can be implemented, some of the above components may be added, changed, or excluded. For example, the embodiments may be realized on a computer network in which a plurality of computers are connected. Further, the type of CPU is arbitrary, and a plurality of CPUs may be used at the same time, or a single CPU may be used for time sharing (time sharing) to perform a plurality of processes simultaneously in parallel.

In addition, other input devices (for example, pointing devices such as touch panels, light pens, trackballs, image input devices such as digitizers, image reading devices and video cameras, voice recognition devices, various sensors, etc.) are used. Good. Also, other external storage devices (for example, floppy disk device, RAM card device, magnetic tape device, optical disk device, magneto-optical disk device,
Bubble memory device, flash memory, etc.) may be used. In addition, other output devices (for example, liquid crystal display device, plasma display device, video projector, LE
D display device, sound generation circuit, voice synthesis circuit, etc.) may be used.

As a software configuration for realizing the embodiment in the computer, typically, application software for realizing each function of the embodiment is executed on an OS (operating system). Embodiments are possible. Further, as a form of software for implementing the embodiment, typically, a machine language compiled (translated) from a high-level language or an assembler can be considered. However, the software configuration of the computer is also free, and the software configuration may be changed as long as the present invention can be implemented. For example, it is not always necessary to use the OS, the expression format of the software is free, and an interpreter (sequential interpretation execution type) language such as BASIC may be used.

The software can be stored in any manner, and may be stored in a ROM (read-only memory). Alternatively, the software may be stored in an external storage device such as a hard disk drive at the time of starting the computer. It may be loaded into the main memory at the start of processing. Alternatively, the software may be divided into a plurality of parts and stored in an external storage device, and only the necessary modules may be loaded (read) into the main memory at any time according to the processing content. Further, each software portion may be stored in a different manner.

Also, each step of each procedure in this embodiment may be executed in a different order, a plurality of steps may be executed at the same time, or may be executed in a different order for each execution, as long as the nature thereof is not violated. Such an order change can be realized by a menu-type interface method, such as a user selecting an executable process.

The term "input" used in this specification includes not only the original input of information but also other processing closely related to the input of information. Such processing is, for example, echo back or correction / editing of input contents. In addition, “input” is not necessarily limited to input from outside the computer, and is also performed by reading information from a predetermined memory area or disk device. Further, in this specification, the term “output” includes not only the original output of information but also other processing closely related to the output of information. Such processing is, for example, input of a range to be output or an instruction for screen scrolling. Also, "output" is
The output is not necessarily limited to the outside of the computer, and may be performed by writing information in a predetermined memory area or a disk device. It should be noted that input and output may be realized by an integrated operation by an interactive input / output procedure, and processing such as selection, designation, and specification may be performed by such an integrated operation.

The data (information) and the data storage means in the present specification may exist in any form on the computer. For example, the location portion of data on the hardware is the main storage device / external storage device / CPU.
It may be any part such as a register or a cache memory. Further, the data retention mode is also free. For example, the data is not only retained in the file format, but may be realized by directly accessing a storage device such as a memory or a disk by a physical address. In addition, the expression format of the data is arbitrary, and for example, the unit of the code representing the character string may be a character unit or a word unit. Further, it is sufficient that the data is retained for the required fixed time, and the data may be deleted thereafter, and the retention time is free. Information that is not changed for the time being, such as dictionary data, may be stored in the ROM.

In this specification, reference to specific information is a confirmation and does not deny the existence of information that is not referred to. That is, in the operation of the present invention, general information necessary for the operation and its storage area, for example, various pointers, stacks, counters, flags, parameters, work areas, buffers, etc. are appropriately used.

The information required by each part of the embodiment for processing is obtained from the other part holding the information unless otherwise specified. Such acquisition of information can be realized, for example, by accessing a variable or memory that stores the information. The information can be erased / erased by not actually deleting the content itself of the information from the storage area but by changing the meaning of the information, such as setting a flag indicating the deletion.

(1) Structure of the First Embodiment The first embodiment has the features of claims 1, 3 to 5, 7, 13, and 15 to 1.
The image processing apparatus corresponds to Nos. 7 and 19 and detects a pixel having a density gradient vector of a desired azimuth angle from each pixel having an intensity value, which constitutes an image on the x and y planes. An object of the first embodiment is to provide an image processing apparatus and an image processing method for efficiently extracting pixels having a density gradient vector of a desired azimuth angle. Another object of the first embodiment is to provide an image processing apparatus and an image processing method in which processing such as confirmation and processing of a detection result is easy. Further, another object of the first embodiment is to provide an image processing apparatus and an image processing method that facilitate the processing suitable for the purpose.

First, FIG. 1 is a functional block diagram showing the configuration of an image processing apparatus (hereinafter referred to as "this apparatus") of the first embodiment. As shown in this figure, the present apparatus designates a detection range that is a range of azimuth angles of the concentration gradient vector to be detected and can be expressed as a range sandwiched between the first azimuth angle and the second azimuth angle. Means 1 and two auxiliary vectors, each orthogonal to the first and second azimuth angles, respectively x
And an auxiliary vector setting means 2 for setting the y component and the y component. Further, the present apparatus further includes a density gradient calculating means 3 for calculating, for each pixel, a density gradient vector as component differential values Gx, Gy of x and y axes based on gray values of the pixel concerned and surrounding pixels. , The inner product means 4 for calculating each inner product value of the density gradient vector and the two auxiliary vectors for each pixel, and the azimuth of the density gradient vector of the pixel is included in the detection range based on the sign of each inner product value. And a determining unit 5 for determining whether or not to perform.

The present apparatus further comprises a first storage means 6 for sequentially storing the detected address value of the pixel on the memory, and a first storage means 6 for converting the address value of the pixel on the memory into coordinate values on the image of the pixel. And the conversion means 7 of 1.

(2) Operation and effect of the first embodiment [Designation of detection range] Pixels Xi, j (i = 0 to N-1, j =, which form an image on the x and y planes and have gray values respectively 0
˜M−1) exists, and the grayscale value of each pixel is represented with an arbitrary precision of a plurality of bits (for example, 8 bits).
In this case, in order to detect a pixel having a density gradient vector of a desired azimuth angle, first, the detection range is designated by the designating means 1. The method of this designation is arbitrary. For example, the user may manually input or may hand over from the processing routine of the application program.

The detection range is a range of azimuth angles of the concentration gradient vector to be detected, and may be expressed as a range sandwiched between the first azimuth angle and the second azimuth angle. That is, here, the detection range is given by the central azimuth angle A [radian] and the detection allowable angle d [radian] before and after the central azimuth angle A + d.
However, the detection range may be specified by the minimum azimuth angle and the maximum azimuth angle. Note that the azimuth angle A-d
Is the first azimuth angle and azimuth angle A + d is the second azimuth angle (FIG. 2).

In the first embodiment, a density gradient vector (Gx, Gy) which has independent two-way differential values as components in each grayscale image is considered. In this case, the azimuth angle tan of the density gradient vector is considered. -1 (Gy / Gx) is the following formula

[Equation 4] Only pixels within the range represented by are extracted.

[Setting of Auxiliary Vector] When the detection range is set, the auxiliary vector setting means 2 causes the first azimuth angle A-
d and two auxiliary vectors S1 and S2, which are respectively orthogonal to the second azimuth angle A + d, are divided into S1 (S1x, S1y), S2 (S2x, S2) by an x component and ay component
Set as in y). FIG. 3 schematically shows two auxiliary vectors set when the detection azimuth angle A and the detection allowable angle d are given as in FIG. Further, FIG. 4 shows each component of the auxiliary vector. The following equation is an example of an operation for calculating each component of the auxiliary vector.

(Equation 5) At this time, each component value S1 of each auxiliary vector S1, S2
x, S1y, S2x, and S2y are each represented by a predetermined range represented by a plurality of bits, that is, an integer value between -L and L. This integer conversion can be realized by conversion such as rounding down or rounding down after the decimal point.

Assuming that A = π / 3 [radian] (= 60
[Degrees]) and d = π / 36 [radians] (= 5 [degrees]), let L be a natural number 256, and make sure that all components of each auxiliary vector take integer values between −L and L. Suppose that it is an integer. In this case, for example, the X component of the auxiliary vector S1 is

(Equation 6) Takes an integer value "-209".

As described above, in the first embodiment, each component value of each auxiliary vector is represented by an integer value within a predetermined range. In addition, since the integer value is generally simple in internal representation format and easy and fast in arithmetic processing, the processing is speeded up (claims 4 and 16).

[Calculation of Density Gradient Vector] Further, in order to detect a pixel, the density gradient calculating means 3 calculates a density gradient vector x, for each pixel based on each gray value of the pixel concerned and surrounding pixels. It is calculated as each component differential value Gx, Gy on the y-axis. It should be noted that this calculation may be performed for all pixels in advance and then proceed to the process after the inner product.
The calculation of the density gradient vector and the process after the inner product may be alternately performed for each pixel.

Here, it is assumed that the components Gx and Gy of the density gradient vector of all pixels are obtained in advance and stored in a predetermined memory area. The density gradient vector at each pixel is calculated as shown in FIG. 7 as in the prior art.

[Determination by Inner Product] Next, the density gradient vector is read out for each pixel, and the inner product means 4 causes the density gradient vector (Gx, Gy) for each pixel and two auxiliary vectors S1 (S1x, S1y). , S2 (S2x, S2y)
Calculate each inner product value of and.

Then, the judging means 5 judges whether or not the density gradient vector of the pixel is included in the detection range based on the sign of each inner product value. Here, in general, the inner product value between two vectors is positive when the angle between the azimuth angles of each vector is acute and negative when the angle is obtuse. Each auxiliary vector is at right angles to the first azimuth angle A-d and the second azimuth angle A + d. Therefore, for the concentration gradient vector, the inner product value Gx · S1x + Gy with each auxiliary vector
The angular relationship between the first azimuth angle A-d and the second azimuth angle A + d can be specified by the symbols S1y, Gx · S1x + Gy · S1y. For example, when the first auxiliary vector is on the second azimuth side with respect to the first azimuth and the second auxiliary vector is on the first azimuth side with respect to the second azimuth, the inner product When the signs of the values are both positive, the concentration gradient vector exists between the first azimuth angle and the second azimuth angle.

The inner product value between two vectors can be calculated by a simple operation such as the sum of the products of x elements and the products of y elements, and the determination of the right angle should be made based only on simple information such as a sign. You can As described above, in the first embodiment, the detection of the pixel having the density gradient vector of the desired azimuth is performed by calculating the azimuth of the density gradient vector of all images in radian (tan ⁻¹ (Gy / Gx)). Without putting
It can be efficiently performed by simple information and processing (claims 1 and 13).

Further, only the inner product value with one of the auxiliary vectors is first calculated, and the azimuth of the density gradient vector of the pixel corresponds to the auxiliary vector based on the sign of the inner product value. When it is determined to be included in the detection range side with respect to the azimuth angle of, the inner product value with the other auxiliary vector may be calculated (claims 3 and 15). By doing this, if it is found that the value is out of the detection range based on the inner product value calculated earlier, the subsequent calculation of the inner product value can be omitted, and the number of calculations can be saved, thus speeding up the processing. To be done.

FIG. 5 is a flow chart showing the procedure of pixel extraction in this case. That is, the density gradient vector (Gx, Gy) for one pixel is read from the memory while moving the pointer indicating the pixel coordinates (step 200),
The inner product P1 with the auxiliary vector S1

(7) P1 = S1x · Gx + S1y · Gy is calculated (step 201). Then, it is determined whether this P1 is positive (P1> 0) (step 20).
2) If it is not positive, the process returns to the movement of the pixel coordinate pointer (step 200) to proceed to the next pixel.

If P1 is positive in step 202, then the inner product P2 of the concentration gradient vector (Gx, Gy) and the auxiliary vector S2 is calculated.

(8) P2 = S2x · Gx + S2y · Gy is calculated (step 203). Then, it is determined whether this P2 is positive (P2> 0) (step 20).
4) If it is not positive, return to the movement of the pixel coordinate pointer (step 200) and proceed to the next pixel. If P1 is positive, the address of the memory in which the grayscale value of the pixel is stored is stored in the first storage means 6 (step 20).
5). If all pixels have not been scanned (step 206), the process proceeds to the next pixel (step 20).
0), if scanning of all pixels has been completed, pixel extraction is completed.

In the first embodiment, not the coordinate value but the address value on the memory is sequentially stored and accumulated for the detected pixel. Therefore, when accessing the gray value of the detected pixel, the coordinate of the image is accessed. The process of converting a value into an address value can be omitted. Therefore, processing such as access to the image data after detection is speeded up (claims 5 and 17).

[Conversion of Extraction Result] As described above, in the first embodiment, the detected portion is stored in the form of the address value on the memory, but the first converting means 7 may store it as necessary. The address value is converted into the coordinate value on the image of the pixel (claims 7 and 19). This conversion can be performed based on the offset value in the memory space of the image memory and the number of pixels in each of the X and Y directions of the image data. For example,
Image memory with 1 byte for 1 pixel is offset address V
And arranged in N pixels in the X direction and M pixels in the Y direction,
When the coordinate value is counted from 0, i and j of the coordinate value Xi, j are respectively calculated based on the address value R.

## EQU9 ## j = int ((RV) / N) i = R- (N * j) -V (where int () is a function that takes the integer part of the argument). As described above, in the first embodiment, the coordinate value on the image of the detected portion can be obtained, so that the processing using the coordinates such as imaging the detected portion and specifying the position is easy.

(3) Second Embodiment Further, the density gradient vector (Gx, Gy) of each pixel and each auxiliary vector (S1x, S1y), (S2x, S2y).
If the parallel processing means for parallel processing the calculation of the inner product value of each of and is provided, the detection processing can be speeded up.
14). Such parallel processing can be realized by a multi-CPU, a numerical operation coprocessor, or the like.

That is, FIG. 6 is a flowchart showing the processing procedure in the second embodiment. As shown in this figure, in the second embodiment, with respect to the density gradient vector of each pixel (step 300), the calculation for obtaining the inner product P1 with the first auxiliary vector S1 and the inner product P with the second auxiliary vector S2 are performed.
The calculation for obtaining 2 is processed in parallel (steps 301, 3
02).

Then, the obtained two inner product values P1 and P2
Are both positive (step 303), and only when both are positive, the address of the pixel is recorded in the extraction memory (step 305).

(4) Other Embodiments The present invention is not limited to the above embodiments,
Since the embodiment can be freely changed, the present invention also includes other embodiments as illustrated below.

For example, the first azimuth angle and the second azimuth angle of the detection range
There are two possible positions at right angles to each azimuth angle, but the auxiliary vector may be set at any right angle position. In either case of setting at right angles, the determination by the sign can be made by the positive or negative sign corresponding to the position.

Further, the means for storing the detection result is not always necessary, and whenever a pixel having a density gradient vector of an azimuth angle within the detection range is detected, a necessary process may be performed on the pixel. Further, the component value of the auxiliary vector does not necessarily have to be represented by an integer, and it is suitable for processing that requires strict accuracy if represented by a real number. Further, each weighting means for weighting the extracted pixels does not necessarily have to be provided.

The image to be processed by the present invention is not limited to an image consisting of only grayscale values, and may be any type of image such as a color image in RGB format or an image based on ultraviolet rays or infrared rays. Can be applied. Further, it is not always necessary to detect pixels for all pixels, and pixels may be thinned out by checking every few pixels. In this case, it is desirable to perform the filtering process in advance so that the detection target pixel appropriately represents the surrounding pixels.

As the detection result, the coordinate values of the detected pixels on the image may be sequentially stored (claims 6 and 1).
8). By doing this, even when performing processing using coordinates such as imaging of the detected portion or specifying the position, the processing of converting the address value on the memory into the coordinate value on the image can be omitted.

Further, the coordinate value of the pixel on the image may be converted into the address value of the pixel on the memory (claims 8 and 20). For example, when an image memory in which 1 pixel is 1 byte is arranged from an offset address V and is configured to have N pixels in the X direction and M pixels in the Y direction and the coordinate value is counted from 0, based on i, j of the coordinate value Xi, j. The address value R is

## EQU10 ## It can be obtained by R = V + j * N + i. By doing this, even if the detected portion is stored in the form of the coordinate value on the image, this coordinate value is converted into the address value in the memory, so the gray value of the detected portion is accessed. For example,
Processing using the address value is easy.

Further, if the detected pixel portion is stored as the image type (claims 9 and 21), the image forming process becomes unnecessary.

Further, each extracted pixel may be weighted based on the magnitude of the gray value (claim 10).
22). Such weighting can be performed in binary depending on whether the gray value exceeds a predetermined threshold value, or by the function value of various functions using the gray value as an argument.
This makes it easy to perform pixel-oriented processing according to the purpose of the processing when the grayscale value has meaning, such as when the dark portion of the image is not the object to be processed but the bright portion is emphasized. Become.

Similarly, the extracted pixels may be weighted based on the magnitude of the density gradient (claims 11 and 23). With this configuration, when the density gradient such as the contour portion of the image has a meaning, it becomes easy to perform the processing that emphasizes the pixels according to the purpose of the processing.

Similarly, each extracted pixel may be weighted based on the difference between the two inner product values of that pixel (claims 12 and 24). That is, generally, the smaller the difference between the inner product values, the closer the density gradient vector of the pixel and the central azimuth angle of the detection range. For this reason, if it makes it above, it will become easy to carry out the processing suitable for the purpose, giving priority to the pixel having the density gradient vector located in the center of the detection range.

[0094]

As described above, according to the present invention, a pixel having a density gradient vector of a desired azimuth can be efficiently extracted, so that image processing can be speeded up.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing the configuration of an image processing apparatus according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a detection range designated in the first embodiment of the present invention.

FIG. 3 is a diagram showing a relationship between a detection range and two auxiliary vectors in the first embodiment of the present invention.

FIG. 4 is a diagram showing the component value of each auxiliary vector in the first embodiment of the present invention.

FIG. 5 is a flowchart showing a detection procedure in the first embodiment of the present invention.

FIG. 6 is a flowchart showing a detection procedure in the second embodiment of the present invention.

FIG. 7 is a principle diagram showing a specific example of calculation of a density gradient vector in the image processing apparatus and the image processing method according to the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Designation means 2 ... Auxiliary vector setting means 3 ... Concentration gradient calculation means 4 ... Inner product means 5 ... Judgment means 6 ... 1st storage means 7 ... 1st conversion means 200 and after ... Each step of a procedure

Claims (24)

[Claims] 1. An image processing apparatus for detecting a pixel having a density gradient vector of a desired azimuth angle from each pixel having an intensity value and forming an image on an x, y plane, and an azimuth of a density gradient vector to be detected. The range of angles, first
Specifying means for specifying a detection range that can be expressed as a range sandwiched between the azimuth angle and the second azimuth angle, and two auxiliary vectors that are respectively orthogonal to the first azimuth angle and the second azimuth angle. With an x-component and a y-component, and a density gradient vector for each pixel, based on the gray values of the pixel and surrounding pixels, for each pixel.
x and Gy, a density gradient calculating means, an inner product means for calculating each inner product value of the density gradient vector and the two auxiliary vectors for each pixel, and an inner product value of each pixel based on the sign of each inner product value. An image processing apparatus comprising: a determination unit that determines whether the azimuth angle of the density gradient vector is included in the detection range. 2. The image processing apparatus according to claim 1, further comprising a parallel processing unit for performing parallel processing for calculating two inner product values for each auxiliary vector. 3. The inner product means first calculates only an inner product value of any one of the auxiliary vectors of the density gradient vector of the pixel, and the density gradient of the pixel is calculated based on the sign of the inner product value. The first direction in which the azimuth angle of the vector corresponds to the auxiliary vector
Alternatively, when it is determined by the determination means that the second azimuth is included in the detection range side, the inner product value with the other auxiliary vector is calculated. 1. The image processing device according to 1. 4. The image processing apparatus according to claim 1, wherein each component value of each auxiliary vector is configured to be represented by an integer value in a predetermined range represented by a plurality of bits. 5. The image processing apparatus according to claim 1, further comprising a first storage unit that sequentially stores the detected address value on the memory of the pixel. 6. The image processing apparatus according to claim 1, further comprising a second storage unit that sequentially stores the detected coordinate values of the pixel on the image. 7. The image processing apparatus according to claim 1, further comprising first conversion means for converting an address value of a pixel on a memory into a coordinate value of the pixel on an image. 8. The image processing apparatus according to claim 1, further comprising a second conversion unit that converts a coordinate value on the image of the pixel into an address value on the memory of the pixel. 9. The image processing apparatus according to claim 1, further comprising a third storage unit for storing an image of the detected pixel portion. 10. The image processing apparatus according to claim 1, further comprising a first weighting unit that weights each extracted pixel based on the magnitude of the grayscale value. 11. The image processing apparatus according to claim 1, further comprising second weighting means for weighting each extracted pixel based on the magnitude of the density gradient vector. 12. A third weighting method for weighting each extracted pixel based on a difference between two inner product values of each pixel.
The image processing apparatus according to claim 1, further comprising: 13. An image processing method for detecting a pixel having a density gradient vector of a desired azimuth angle from pixels each of which has an intensity value and constitutes an image on an x, y plane, and an azimuth of a density gradient vector to be detected. The range of angles, first
A step of designating a detection range that can be expressed as a range sandwiched between the azimuth angle and the second azimuth angle, and two auxiliary angles each orthogonal to the first azimuth angle and the second azimuth angle. Auxiliary vector setting step of setting vector by x component and y component respectively, and for each pixel, based on each gray value of the pixel concerned and surrounding pixels, the density gradient vector G
x and Gy, a step of calculating a density gradient, a step of calculating an inner product value of each of the density gradient vector and the two auxiliary vectors for each pixel, and a step of calculating the inner product value based on the sign of each inner product value. A step of determining whether or not the azimuth angle of the density gradient vector of the pixel is included in the detection range. 14. The image processing method according to claim 13, further comprising a step of parallel processing for parallel processing of calculating two inner product values for each auxiliary vector. 15. The inner product step first calculates only an inner product value of the density gradient vector of the pixel and one of the auxiliary vectors, and the density of the pixel is calculated based on a sign of the inner product value. The first direction in which the azimuth angle of the gradient vector corresponds to the auxiliary vector
14. The image according to claim 13, wherein when the determination step determines that the second azimuth angle is included in the detection range side, an inner product value with the other auxiliary vector is calculated. Processing method. 16. The image processing method according to claim 13, wherein each component value of each auxiliary vector is represented by an integer value in a predetermined range represented by a plurality of bits. 17. The image processing method according to claim 13, further comprising a first storage step of sequentially storing the detected address value of the pixel on the memory. 18. The image processing method according to claim 13, further comprising a second storage step of sequentially storing the detected coordinate values of the pixel on the image. 19. The image processing method according to claim 13, further comprising a first conversion step of converting an address value of a pixel on a memory into a coordinate value on an image of the pixel. 20. The image processing method according to claim 13, further comprising a second conversion step of converting the coordinate value of the pixel on the image into the address value of the pixel on the memory. 21. The image processing method according to claim 13, further comprising a third storing step of storing an image of the detected pixel portion. 22. The image processing method according to claim 13, further comprising a first weighting step of weighting each extracted pixel based on the magnitude of the gray value. 23. The image processing method according to claim 13, further comprising a second weighting step of weighting each extracted pixel based on the magnitude of the density gradient vector. 24. A third weighting method for weighting each extracted pixel based on a difference between two inner product values of each pixel.
2. The method of claim 1 including the step of
3. The image processing method described in 3.