WO2013161111A1 - 画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム - Google Patents
画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム Download PDFInfo
- Publication number
- WO2013161111A1 WO2013161111A1 PCT/JP2012/079702 JP2012079702W WO2013161111A1 WO 2013161111 A1 WO2013161111 A1 WO 2013161111A1 JP 2012079702 W JP2012079702 W JP 2012079702W WO 2013161111 A1 WO2013161111 A1 WO 2013161111A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- pixel
- difference
- unit
- roughness
- Prior art date
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 72
- 238000004364 calculation method Methods 0.000 claims abstract description 20
- 238000000034 method Methods 0.000 claims description 41
- 230000006870 function Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 8
- 238000009499 grossing Methods 0.000 description 8
- 230000002093 peripheral effect Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- GZPBVLUEICLBOA-UHFFFAOYSA-N 4-(dimethylamino)-3,5-dimethylphenol Chemical compound CN(C)C1=C(C)C=C(O)C=C1C GZPBVLUEICLBOA-UHFFFAOYSA-N 0.000 description 5
- 238000012935 Averaging Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000010187 selection method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20224—Image subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Definitions
- the present invention relates to an image evaluation device, an image selection device, an image evaluation method, a recording medium, and a program, and appropriately evaluates the roughness of the image.
- Patent Document 1 proposes a technique that objectively evaluates an evaluation target image obtained by performing image processing on an original image using entropy.
- an original image before image processing is required as a comparison image in addition to an evaluation target image.
- Non-Patent Document 1 research is also proceeding on pseudo Hilbert scanning that passes through all the pixels of the rectangular image and satisfies the condition that the pixels that are close in the path are also close in the rectangular image. Applications such as image compression are being promoted.
- the present invention solves the above-described problems, and an object thereof is to provide an image evaluation device, an image selection device, an image evaluation method, a recording medium, and a program for appropriately evaluating the roughness of an image. .
- An image evaluation apparatus includes: A blurring unit for generating a second image obtained by blurring the first image; A different portion for generating a third image representing a difference in pixel value of each pixel between the first image and the second image; A scanning unit that scans each pixel included in the third image to obtain a difference between pixel values of adjacent pixels, and obtains an appearance probability for each of the obtained differences; A calculation unit for calculating entropy from the occurrence probabilities for each obtained difference; An output unit that outputs the entropy is provided as an evaluation value of the roughness of the first image.
- the scanning unit may be configured to scan each pixel included in the third image from left to right and from top to bottom.
- the scanning unit may be configured to scan each pixel included in the third image along a space filling curve.
- the pixel value of each pixel included in the third image is a pixel value at the position of the pixel in the first image in a predetermined color space and a pixel value at the position of the pixel in the second image. It can be configured to be a distance.
- the difference between the pixel values of the adjacent pixels can be configured to be the distance between the pixel values of the adjacent pixels in a predetermined color space.
- An image selection device provides: A reception unit for receiving a plurality of images depicting one target; An acquisition unit that acquires the evaluation value of the roughness of each of the plurality of received images by the image evaluation device, A selection unit that selects an image with the lowest roughness from the received plurality of images based on the acquired evaluation value of the roughness is provided.
- An image evaluation method includes: An image evaluation apparatus having a blur part, a difference part, a scanning part, a calculation part, and an output part is executed, A blurring step in which the blurring unit generates a second image obtained by blurring the first image; A different step in which the different part generates a third image representing a difference in pixel values of the pixels of the first image and the second image; A scanning step in which the scanning unit scans each pixel included in the third image to obtain a difference between pixel values of adjacent pixels, and obtains an appearance probability for each of the obtained differences; A calculation step in which the calculation unit calculates entropy from the occurrence probability for each of the obtained differences; The output unit includes an output step of outputting the entropy as an evaluation value of the roughness of the first image.
- a computer-readable recording medium is provided.
- Computer A blurring unit for generating a second image obtained by blurring the first image; A different portion for generating a third image representing a difference in pixel value of each pixel between the first image and the second image; A scanning unit that scans each pixel included in the third image to obtain a difference between pixel values of adjacent pixels, and obtains an appearance probability for each of the obtained differences; A calculation unit for calculating entropy from the occurrence probabilities for each obtained difference; A program that functions as an output unit that outputs the entropy is recorded as an evaluation value of the roughness of the first image.
- a program provides a computer, A blurring unit for generating a second image obtained by blurring the first image; A different portion for generating a third image representing a difference in pixel value of each pixel between the first image and the second image; A scanning unit that scans each pixel included in the third image to obtain a difference between pixel values of adjacent pixels, and obtains an appearance probability for each of the obtained differences; A calculation unit for calculating entropy from the occurrence probabilities for each obtained difference; The evaluation value of the roughness of the first image is configured to function as an output unit that outputs the entropy.
- the program of the present invention can be recorded on a computer-readable non-transitory recording medium such as a compact disk, flexible disk, hard disk, magneto-optical disk, digital video disk, magnetic tape, and semiconductor memory. . These recording media can be distributed and sold independently of the computer.
- the program of the present invention is loaded from a recording medium as described above onto a computer readable / writable recording medium such as a RAM (Random Access Memory), temporarily recorded, and then stored in a CPU (Central Processing Unit) can be configured to read, interpret, and execute a program recorded in the RAM or the like.
- a computer readable / writable recording medium such as a RAM (Random Access Memory)
- a CPU Central Processing Unit
- the program of the present invention can be distributed and sold via a transitory transmission medium such as a computer communication network, independently of the computer on which the program is executed.
- an image evaluation device it is possible to provide an image evaluation device, an image selection device, an image evaluation method, a recording medium, and a program that appropriately evaluate the roughness of an image.
- the image evaluation apparatus can be realized by executing a predetermined program on various computers such as a server computer and a personal computer.
- the computer uses the RAM as a temporary storage area or an output destination of the processing result when the CPU executes a program, receives an instruction from a user by an input device such as a keyboard or a mouse,
- This is hardware that outputs the result of processing to an output device such as a display or performs the above input / output by communicating with other devices via a NIC (Network Interface Card). It can be omitted as appropriate.
- images to be processed by the CPU are recorded on the hard disk of the computer.
- images are managed by a file system and various databases together with information about a photographer, a shooting date and time, a shooting location, and a shooting target.
- the image evaluation apparatus is realized by a plurality of computers connected via a dedicated communication line, a communication line, or a computer communication network such as the Internet, executing the above processing in parallel, distributed, or in parallel. It is also possible to speed up the process.
- an electronic circuit design is created from a program, and a dedicated electronic circuit is configured based on the design, thereby evaluating the image of the present invention. It is also possible to implement the device.
- FPGA Field Programmable Gate Array
- the image to be evaluated is obtained as a result of photographing a real object with a digital camera, or as a result of scanning a film or paper with a scanner, and is an image that can be digitally processed.
- the image a has a rectangular shape, has a width a.W dots, a height a.H dots, and is expressed as a set of a.W ⁇ a.H pixels.
- the position of each pixel will be referred to in the following order from the upper left. That is, the first element of coordinates representing pixel positions is defined by the horizontal axis from left to right, and the second element of the coordinates is defined by the vertical axis from top to bottom.
- the pixel value of the pixel at the coordinates (x, y) included in the image a is expressed as a [x, y].
- Each pixel value is represented by a three-dimensional vector of scalar values for monochrome images and red, green, and blue for color images.
- digital representation is used.
- the pixel value takes an integer value from 0 to 255
- three pixel values of red, green, and blue are used. Each takes an integer value from 0 to 255.
- the absolute value of the difference between the scalar values of the pixel values can be adopted for a monochrome image.
- the pixel value vector distance (the square root of the sum of squares of the differences between the red, green, and blue elements), the square distance (the square sum of the differences of the red, green, and blue elements), and the Manhattan distance (red , The sum of absolute values of differences between green and blue).
- a method of using an image obtained by making a color image monochrome as a distance between two pixel values included in the color image That is, the absolute value of the difference between the pixel values (scalar values) of two monochrome pixels obtained by converting the two color pixels into monochrome is adopted as the distance between the pixel values of the two color pixels.
- each pixel included in the image is scanned. That is, processing for arranging pixels included in an image in a row is executed.
- FIG. 1A, FIG. 1B, and FIG. 1C are explanatory diagrams illustrating paths for scanning pixels. Hereinafter, description will be given with reference to these drawings.
- an image 11 composed of 8 ⁇ 8 pixels is scanned by paths 12a, 12b, and 12c.
- the coordinates of the i-th pixel of the image a are (i mod a.W, i div a.W).
- x div y means integer division (result of dividing x by y)
- x mod y means a remainder in integer division (remainder of dividing x by y).
- the second method is to calculate the coordinates of the i-th pixel of image a, (A) If i div (aW ⁇ 2) is an even number, (i mod aW, i div aW) (B) If i div (aW ⁇ 2) is an odd number, (aW ⁇ (i mod aW) ⁇ 1, i div aW) is set.
- each pixel of the image 11 is scanned along the path 12a. That is, first, the horizontal scanning is performed from the left end to the right end, and when the right end is reached, the scanning is performed vertically downward by 1 dot, the horizontal scanning is performed from the right end to the left end, and when the left end is reached, the scanning is performed downward by 1 dot. It moves vertically and repeats scanning horizontally from the left end to the right end.
- the third method is a method using pseudo Hilbert scanning as disclosed in Non-Patent Document 1.
- the second method pixels vertically adjacent in the image are separated during scanning, but in this method, the pixels included in the image 11 are adjacent to each other as shown in FIG. 1C. If they are aligned or adjacent to each other in the vertical direction, the degree and frequency of proximity in the scanning path 12c will be greater than in the case of FIG. 1B.
- the pixel value of the i-th pixel in the path 12 in the scanning of the image 11 is represented as a (i).
- the column of pixel values a (0), a (1), a (2), ..., a (a.W ⁇ a.H-1) obtained by scanning are the following pixel values:
- FIG. 2 is an explanatory diagram showing a schematic configuration of the image evaluation apparatus according to the present embodiment
- FIG. 3 is a flowchart showing a control flow of image evaluation processing executed by the image evaluation apparatus.
- the image evaluation apparatus 101 includes a blur unit 102, a difference unit 103, a scanning unit 104, a calculation unit 105, and an output unit 106, as shown in FIG.
- the functions of these units are realized by a computer executing a predetermined program.
- the image evaluation apparatus 101 will be described on the assumption that the first image a is an evaluation target.
- any one of the above-described various scanning paths is selected and applied to all the images to be processed.
- the information of the first image a and all the images to be processed later will be recorded on a recording medium such as a hard disk, transmitted and received via a computer communication network, or temporarily stored in a RAM or the like. By storing in the area, various processes can be performed.
- the blurring unit 102 generates a second image b in which the first image a is blurred (step S301).
- the blur processing can be realized by applying a so-called smoothing filter to the image.
- “ ⁇ ” means substitution.
- two-dimensional smoothing can also be performed.
- two-dimensional smoothing can also be performed.
- b [x, y] ⁇ [a [x-1, y] + a [x + 1, y] + a [x, y-1] + a [x, y + 1] + B ⁇ b [x, y]] / (B + 4)
- smoothing is performed by taking a weighted average of a certain pixel and its upper, lower, left and right adjacent pixels.
- blurring is performed by taking the average of the pixel value at a desired position and the pixel value at the next position.
- the desired position is the last pixel of the second image b
- the next position is outside the range of the first image a. Therefore, in this case, the pixel value of the pixel at the last position of the first image a is used as it is.
- FIG. 4A is an explanatory diagram showing the state of pixel values obtained when the first image a composed of a rough image depicting this object is scanned.
- FIG. 4B is an explanatory diagram showing the state of pixel values obtained when the first image a composed of a smooth image depicting a certain object is scanned.
- FIG. 5A is an explanatory diagram illustrating a state of pixel values obtained when the second image b with respect to the first image a illustrated in FIG. 4A is scanned.
- FIG. 5B is an explanatory diagram illustrating a state of pixel values obtained when the second image b with respect to the first image a illustrated in FIG. 4B is scanned.
- the horizontal axis represents the order in the scanning path
- the vertical axis represents the pixel value of the pixels in the order.
- a case where the pixel value is a scalar value is shown.
- the pixel value of the first image a changes smoothly, but there are peaks protruding in some places and depressions in the bottom.
- FIG. 5A is a pixel value of the second image b obtained by smoothing the coarse first image a according to FIG. 4A
- FIG. 5B is a diagram of the second image b obtained by smoothing the smooth first image a according to FIG. 4B. It is a pixel value.
- 5A and 5B is similar to the change in the first image a in FIGS. 4A and 4B when viewed as a whole.
- the addition / subtraction / division of the pixel value is based on the same calculation method as the addition / subtraction / division of a normal scalar value or vector value. However, if each element is normalized to an integer value, the sizes of the first image a and the second image b are the same. Can be sized.
- the different unit 103 generates a third image c representing the difference between the first image a and the second image b (step S302).
- the difference between the first image a and the second image b is expressed by the distance between the pixel values arranged at the same position in the image.
- the third image c has the same vertical and horizontal sizes as the first image a and the second image b.
- the third image c can be generated.
- the pixels may be operated based on other orders. For example, c [x, y] ⁇
- the operation may be performed along a row or a column of pixels in the image.
- FIGS. 6A and 6B show the state of the pixel values of the first image a shown in FIGS. 4A and 4B and the state of the pixel values of the second image b shown in FIGS. 5A and 5B. It is explanatory drawing showing the mode of a pixel value.
- FIG. 6A and 6B show the state of the pixel values of the first image a shown in FIGS. 4A and 4B and the state of the pixel values of the second image b shown in FIGS. 5A and 5B.
- the third image c shown in FIG. 6B is generated from the smooth first image a, and the peak portion and the bottom portion corresponding to noise hardly appear and the pixel value is 0 or a value close to 0. .
- the third image c shown in FIG. 6A is generated from the coarse first image a, and a peak portion and a bottom portion corresponding to noise protrude and remain.
- the pixel value is 0 or a value close to 0 in a portion without noise.
- entropy is used in the present invention in order to set the degree of such protrusion and remaining as an evaluation value.
- the scanning unit 104 scans each pixel included in the third image c, and obtains the appearance probability of the difference between the pixel values of adjacent pixels (step S303).
- each pixel value may be a one-dimensional array or an associative array.
- each element of the array t is accessed like t [x].
- the number of elements may be a value obtained by adding 1 to the maximum value DMAX that can be a distance between pixel values.
- the array t may be an associative array using a hash or the like. Since most of the pixel value differences are extremely close to 0, the memory can be used efficiently by using a hash or the like. In this case, the setting that the default value of each element of the array t is 0 may be executed.
- each element t [x] of the array t needs to be able to store the maximum pixel value distance.
- each pixel value of the first image a, the second image b, and the third image c can be expressed by an integer value between 0 and 255. Therefore, a 1-byte area may be secured for each element of the array representing these.
- the image size can have various sizes, but the widely used image size can be expressed by 2-byte integers in both width and height. Therefore, it is sufficient to secure a 4-byte area for each element of the array t.
- the third image c can be viewed as a one-dimensional array when scanned along the path, and corresponds to a two-dimensional array when viewed as a normal image representation. Therefore, it is possible to apply various techniques for enumerating the distribution of values of each element of the array in parallel and in parallel.
- the calculation unit 105 calculates entropy E from the appearance probability p (d) for each difference obtained for each difference d (step S304).
- E - ⁇ d ⁇ keys (t) p (d) ⁇ log (p (d))
- the output unit 106 outputs entropy E as the roughness evaluation value of the first image a (step S305).
- the entropy E calculated as described above is high. Therefore, the entropy E can be used as an index of image roughness.
- Patent Document 1 the evaluation target image is compared with the original image to obtain the roughness of the evaluation target image.
- the roughness of the first image a is obtained only from the first image a to be evaluated.
- the second image b obtained by blurring the first image a and the third image c representing the difference between the first image a and the second image b are sequentially generated from the first image a.
- the appearance frequency of each pixel value in the three images c is counted up. Therefore, high-speed calculation is possible by using a CPU or coprocessor having multimedia computing hardware for obtaining image blurring or difference.
- the pixel value c (i) is the distance between b (i) and a (i), and b (i) can be obtained from the pixel values of the pixels near the i-th pixel of the first image a. it can. Therefore, if each c (i) is directly obtained from the pixel values of necessary pixels of the first image a and the array t is updated with the obtained values, the second image b and the third image c are stored. Therefore, it is possible to reduce the memory usage by omitting the image storage area.
- the peripheral edge of the second image b has a filter that is smaller than the first image a by D dots in the top, bottom, left, and right.
- the vertical and horizontal size of the image does not change before and after the blur by changing the range of the image area in which the averaging calculation is performed.
- the latter two-dimensional blur filter can be used as it is in the above embodiment.
- the former two-dimensional blur filter is applied to the above embodiment,
- the method (a) ignores the peripheral edges when entropy is calculated.
- the pixels corresponding to the peripheral edge D dots are handled as having no difference in pixel values between the first image a and the second image b.
- the background around the edge of the image is displayed and there is often no important information. Further, in the number of pixels for the peripheral edge D dots and the number of other pixels, the latter is generally more in each stage. Therefore, even if the methods (a) and (b) are employed, sufficient results can be obtained as an index for evaluating the roughness of the image.
- the pixel value of each pixel in the third image represents the difference between the pixel values of the pixels arranged at the same position in the first image and the second image.
- the correspondence of the positions can be arbitrarily changed. That is, the difference in the pixel value at each position of the first image and the second image can be expressed by the pixel at the same position in the third image, but does not necessarily need to match.
- the difference between the pixels at each position in the first image and the second image can be stored in the pixel at the mirror image position or the inverted position at each position in the third image.
- the pixel value of each pixel in the third image represents the difference between the pixel values of the pixels arranged at positions corresponding to the respective pixels in the first image and the second image.
- each shop owner In an electronic shopping mall in which a plurality of shop owners exhibit the same product, each shop owner often prepares an image that captures or depicts the product.
- These images include images prepared in advance by the manufacturer of the product, images of the product itself taken by each store owner with a digital camera, etc., and each of these stores has an advertising phrase, store name, store logo, price There are images that have been modified with information such as
- the image evaluation method according to the above-described embodiment can be used to appropriately select a representative image showing a product when a plurality of images depicting a product are prepared.
- the image selection device performs such image selection.
- the present image selection apparatus can be applied for any purpose when selecting any one of a plurality of images. For example, when a user of a digital camera shoots an object several times in succession, in the case of automatically selecting an image with the best shooting result, or in a moving image configured by arranging a plurality of frame images in order
- the present invention can also be applied to a case where a frame image representing the moving image is selected from the frame images.
- This image selection apparatus can be realized by causing a computer to execute a predetermined program, as with the image evaluation apparatus 101 described above, and can also be implemented as an electronic circuit using an FPGA or the like.
- FIG. 7 is an explanatory diagram showing a schematic configuration of the image selection device according to the present embodiment.
- FIG. 8 is a flowchart showing the flow of processing of the image selection method executed by the image selection apparatus.
- the image selection device 601 includes a reception unit 602, an acquisition unit 603, and a selection unit 604.
- the reception unit 602 first receives a plurality of images depicting one target by starting the program (step S701). ).
- a plurality of images are typically recorded on a hard disk, but a mode of sequentially acquiring from a computer communication network may be adopted.
- the acquisition unit 603 acquires the evaluation value of the roughness of each of the received plurality of images by the image evaluation apparatus 101 (step S702).
- the image evaluation apparatus 101 can independently evaluate the roughness of each image without comparing the original image and other images. Therefore, the evaluation value of the roughness can be calculated in parallel or in parallel.
- the selection unit 604 selects the image with the lowest roughness, that is, the image with the lowest entropy from the received plurality of images based on the acquired evaluation value of the roughness (step S703). The process ends.
- the obtained roughness is the entropy of a non-smooth part in the image.
- the entropy increases when the image contains a lot of noise or when new information (for example, an image of a character string in a promotional phrase) is overwritten on the image.
- an image evaluation device it is possible to provide an image evaluation device, an image selection device, an image evaluation method, a recording medium, and a program that appropriately evaluate the roughness of an image.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
Description
第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
を備えるように構成する。
前記走査部は、前記第3画像に含まれる各画素を、左から右へ、上から下へ、走査する
ように構成することができる。
前記走査部は、前記第3画像に含まれる各画素を、空間充填曲線に沿って走査する
ように構成することができる。
前記第3画像に含まれる各画素の画素値は、所定の色空間における前記第1画像の当該各画素の位置における画素値と、前記第2画像の当該各画素の位置における画素値と、の距離である
ように構成することができる。
前記隣接する画素の画素値の相違は、所定の色空間における当該隣接する画素の画素値の距離である
ように構成することができる。
1つの対象を描写した複数の画像を受け付ける受付部、
前記受け付けられた複数の画像のそれぞれの粗度の評価値を、上記の画像評価装置により、取得する取得部、
前記取得された粗度の評価値により、前記受け付けられた複数の画像から、最も粗度が低い画像を選択する選択部
を備えるように構成する。
ぼかし部、相違部、走査部、計算部、出力部を有する画像評価装置が実行し、
前記ぼかし部が、第1画像をぼかした第2画像を生成するぼかし工程、
前記相違部が、前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違工程、
前記走査部が、前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査工程、
前記計算部が、前記求められた相違ごとの出現確率からエントロピーを計算する計算工程、
前記出力部が、前記第1画像の粗度の評価値として、前記エントロピーを出力する出力工程
を備えるように構成する。
コンピュータを、
第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
として機能させるプログラムを記録するように構成する。
第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
として機能させるように構成する。
本実施形態では、画像の粗度を評価する。評価の対象となる画像は、ディジタルカメラで現実の物体を撮影した結果や、フィルムや紙などをスキャナでスキャンした結果得られるもので、ディジタル処理が可能な画像である。
(a)i div (a.W×2)が偶数であれば、(i mod a.W,i div a.W)とし、
(b)i div (a.W×2)が奇数であれば、(a.W - (i mod a.W) - 1,i div a.W)とする。
の順序を入れ換えたものに相当する。
図2は、本実施形態に係る画像評価装置の概要構成を示す説明図であり、図3は、画像評価装置が実行する画像評価処理の制御の流れを示すフローチャートである。以下、これらの図を参照して説明する。
a.W = b.W,a.H = b.H
が成立する。
b(i) ← 〔a(i)+a(i+1)〕/2, (i = 0,1,...,a.W×a.H-2);
b(i) ← a(i), (i = a.W×a.H-1)
とする手法では、隣りの画素との平均をとることにより、平滑化する。ここで、「←」は代入を意味する。
b(i) ← a(i), (i = 0);
b(i) ← 〔a(i-1)+B×a(i)+a(i+1)〕/(B+2), (i = 1,2,...,W×a.H-2);
b(i) ← a(i), (i = a.W×a.H-1)
とする手法では、経路中の前後の画素との重み付き平均をとることにより、平滑化がなされる。
b[x,y] ←〔a[x-1,y]+a[x+1,y]+a[x,y-1]+a[x,y+1]+B×b[x,y]〕/(B+4),
(x = 1,2,...,a.W-2; y = 1,2,...,a.H-2);
b[x,y] ← a[x,y], (x=0 or x=a.W-1 or y=0 or y=a.H-1)
とする手法では、ある画素と、その上下左右の隣接画素と、の重み付き平均をとることにより、平滑化がなされる。
a.W = b.W = c.W,a.H = b.H = c.H
c(i) ← |b(i)-a(i)|, (i = 0,1,2,...,a.W×a.H-1)
のように第3画像cの画素値を定めることとすれば、第3画像cを生成することができる。ただし、他の順序に基づいて画素を操作することとしても良い。たとえば、
c[x,y] ← |b[x,y] - a[x,y]|,
(x = 0,1,...,a.W-1; y = 0,1,...,a.H-1)
のように、画像内の画素の行や列に沿って操作しても良い。
t[d] ← 0, (d = 0,1,...,DMAX)
を実行する必要がある。
t[c(i)] ← t[c(i)] + 1, (i = 0,1,…,a.W×a.H-1)
p(d) = t[d]/(a.W×a.H)
と計算することができる。
keys(t) = {d | d∈{0,1,...,DMAX},t[d]>0}
E = - Σd∈keys(t) p(d)×log〔p(d)〕
(a)ぼかし以降の処理においては、第1画像aの周囲の縁を、上下左右とも、Dドット分だけ、除去して、第2画像bと同じ大きさとする、あるいは、
(b)第2画像bの周囲の縁のDドット分の画素は、第1画像aと同じ画素値とする、
のいずれかの手法を採用すれば良い。
102 ぼかし部
103 相違部
104 走査部
105 計算部
106 出力部
601 画像選択装置
602 受付部
603 取得部
604 選択部
Claims (9)
- 第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
を備えることを特徴とする画像評価装置。 - 請求項1に記載の画像評価装置であって、
前記走査部は、前記第3画像に含まれる各画素を、左から右へ、上から下へ、走査する
ことを特徴とする画像評価装置。 - 請求項1に記載の画像評価装置であって、
前記走査部は、前記第3画像に含まれる各画素を、空間充填曲線に沿って走査する
ことを特徴とする画像評価装置。 - 請求項1から3のいずれか1項に記載の画像評価装置であって、
前記第3画像に含まれる各画素の画素値は、所定の色空間における前記第1画像の当該各画素の位置における画素値と、前記第2画像の当該各画素の位置における画素値と、の距離である
ことを特徴とする画像評価装置。 - 請求項1から4のいずれか1項に記載の画像評価装置であって、
前記隣接する画素の画素値の相違は、所定の色空間における当該隣接する画素の画素値の距離である
ことを特徴とする画像評価装置。 - 1つの対象を描写した複数の画像を受け付ける受付部、
前記受け付けられた複数の画像のそれぞれの粗度の評価値を、請求項1から5のいずれか1項に記載の画像評価装置により、取得する取得部、
前記取得された粗度の評価値により、前記受け付けられた複数の画像から、最も粗度が低い画像を選択する選択部
を備えることを特徴とする画像選択装置。 - ぼかし部、相違部、走査部、計算部、出力部を有する画像評価装置が実行する画像評価方法であって、
前記ぼかし部が、第1画像をぼかした第2画像を生成するぼかし工程、
前記相違部が、前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違工程、
前記走査部が、前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査工程、
前記計算部が、前記求められた相違ごとの出現確率からエントロピーを計算する計算工程、
前記出力部が、前記第1画像の粗度の評価値として、前記エントロピーを出力する出力工程
を備えることを特徴とする画像評価方法。 - コンピュータを、
第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
として機能させることを特徴とするプログラムを記録したコンピュータ読取可能な記録媒体。 - コンピュータを、
第1画像をぼかした第2画像を生成するぼかし部、
前記第1画像と前記第2画像との各画素の画素値の相違を表す第3画像を生成する相違部、
前記第3画像に含まれる各画素を走査して、隣接する画素の画素値の相違を求め、前記求められた相違ごとの出現確率を求める走査部、
前記求められた相違ごとの出現確率からエントロピーを計算する計算部、
前記第1画像の粗度の評価値として、前記エントロピーを出力する出力部
として機能させることを特徴とするプログラム。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP12875596.4A EP2765555B1 (en) | 2012-04-25 | 2012-11-15 | Image evaluation device, image selection device, image evaluation method, recording medium, and program |
ES12875596.4T ES2575009T3 (es) | 2012-04-25 | 2012-11-15 | Dispositivo de evaluación de imagen, dispositivo de selección de imagen, procedimiento de evaluación de imagen, medio de almacenamiento, y programa |
US14/356,039 US9330447B2 (en) | 2012-04-25 | 2012-11-15 | Image evaluation device, image selection device, image evaluation method, recording medium, and program |
JP2013509350A JP5248719B1 (ja) | 2012-04-25 | 2012-11-15 | 画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2012-100211 | 2012-04-25 | ||
JP2012100211 | 2012-04-25 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2013161111A1 true WO2013161111A1 (ja) | 2013-10-31 |
Family
ID=49482472
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2012/079702 WO2013161111A1 (ja) | 2012-04-25 | 2012-11-15 | 画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム |
Country Status (4)
Country | Link |
---|---|
US (1) | US9330447B2 (ja) |
EP (1) | EP2765555B1 (ja) |
ES (1) | ES2575009T3 (ja) |
WO (1) | WO2013161111A1 (ja) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104677314A (zh) * | 2015-03-02 | 2015-06-03 | 合肥京东方光电科技有限公司 | 检测显示面板表面平坦度的装置及方法 |
US10339410B1 (en) * | 2016-01-13 | 2019-07-02 | Snap Inc. | Color extraction of a video stream |
CN110809885A (zh) * | 2017-06-28 | 2020-02-18 | 高途乐公司 | 图像传感器瑕疵检测 |
US10880092B2 (en) * | 2018-02-05 | 2020-12-29 | Colossio, Inc. | Compression and manipulation-resistant fuzzy hashing |
CN112529831B (zh) * | 2019-08-28 | 2024-05-24 | 深圳市熠摄科技有限公司 | 利用图像处理技术的地貌潜变观测设备 |
CN112526705A (zh) * | 2020-12-07 | 2021-03-19 | 中国科学院长春光学精密机械与物理研究所 | 航天器光学载荷在轨自适应调焦方法、装置、设备及介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10200893A (ja) | 1997-01-14 | 1998-07-31 | Nippon Telegr & Teleph Corp <Ntt> | 画像評価方法 |
JP2004272565A (ja) * | 2003-03-07 | 2004-09-30 | Ricoh Co Ltd | 画像評価装置、画像評価方法、およびその方法をコンピュータが実行するためのプログラム |
JP2009105637A (ja) * | 2007-10-23 | 2009-05-14 | Kyodo Printing Co Ltd | 画像のぼけ判断方法、その装置及びそのプログラム |
JP2011113526A (ja) * | 2009-11-30 | 2011-06-09 | Nikon Corp | 画像判定装置および画像判定プログラム |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5436979A (en) * | 1992-08-21 | 1995-07-25 | Eastman Kodak Company | Process for detecting and mapping dirt on the surface of a photographic element |
US20060008174A1 (en) * | 2004-07-07 | 2006-01-12 | Ge Medical Systems Global Technology | Count adaptive noise reduction method of x-ray images |
WO2006113583A2 (en) * | 2005-04-15 | 2006-10-26 | Mississippi State University | Remote sensing imagery accuracy analysis method and apparatus |
WO2006114003A1 (en) * | 2005-04-27 | 2006-11-02 | The Governors Of The University Of Alberta | A method and system for automatic detection and segmentation of tumors and associated edema (swelling) in magnetic resonance (mri) images |
JP4427001B2 (ja) * | 2005-05-13 | 2010-03-03 | オリンパス株式会社 | 画像処理装置、画像処理プログラム |
US8538102B2 (en) * | 2008-12-17 | 2013-09-17 | Synarc Inc | Optimised region of interest selection |
SE536510C2 (sv) * | 2012-02-21 | 2014-01-14 | Flir Systems Ab | Bildbehandlingsmetod för detaljförstärkning och brusreduktion |
-
2012
- 2012-11-15 WO PCT/JP2012/079702 patent/WO2013161111A1/ja active Application Filing
- 2012-11-15 US US14/356,039 patent/US9330447B2/en active Active
- 2012-11-15 ES ES12875596.4T patent/ES2575009T3/es active Active
- 2012-11-15 EP EP12875596.4A patent/EP2765555B1/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10200893A (ja) | 1997-01-14 | 1998-07-31 | Nippon Telegr & Teleph Corp <Ntt> | 画像評価方法 |
JP2004272565A (ja) * | 2003-03-07 | 2004-09-30 | Ricoh Co Ltd | 画像評価装置、画像評価方法、およびその方法をコンピュータが実行するためのプログラム |
JP2009105637A (ja) * | 2007-10-23 | 2009-05-14 | Kyodo Printing Co Ltd | 画像のぼけ判断方法、その装置及びそのプログラム |
JP2011113526A (ja) * | 2009-11-30 | 2011-06-09 | Nikon Corp | 画像判定装置および画像判定プログラム |
Non-Patent Citations (2)
Title |
---|
JIAN ZHANG; SEI-ICHIRO KAMATA; YOSHIFUMI UESHIGE: "A Pseudo-Hilbert Scan Algorithm for Arbitrarily-Sized Rectangle Region, ADVANCES IN MACHINE VISION, IMAGE PROCESSING, AND PATTERN ANALYSIS", LECTURE NOTES IN COMPUTER SCIENCE, vol. 4153, 2006, pages 290 - 299 |
See also references of EP2765555A4 |
Also Published As
Publication number | Publication date |
---|---|
ES2575009T3 (es) | 2016-06-23 |
EP2765555A1 (en) | 2014-08-13 |
EP2765555B1 (en) | 2016-03-23 |
US20140301640A1 (en) | 2014-10-09 |
US9330447B2 (en) | 2016-05-03 |
EP2765555A4 (en) | 2015-04-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10452905B2 (en) | System and method for detecting objects in an image | |
CN107409166B (zh) | 摇摄镜头的自动生成 | |
WO2013161111A1 (ja) | 画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム | |
CN110827200A (zh) | 一种图像超分重建方法、图像超分重建装置及移动终端 | |
JP4556813B2 (ja) | 画像処理装置、及びプログラム | |
CN108027961A (zh) | 用于逆色调映射的方法和装置 | |
CN111131688B (zh) | 一种图像处理方法、装置及移动终端 | |
KR20110131949A (ko) | 영상 처리 장치 및 방법 | |
CN104853063B (zh) | 一种基于sse2指令集的图像锐化方法 | |
JP2021189527A (ja) | 情報処理装置、情報処理方法及びプログラム | |
EP1963970A2 (fr) | Procede pour fournir des donnees a un moyen de traitement numerique | |
JP5248719B1 (ja) | 画像評価装置、画像選択装置、画像評価方法、記録媒体、ならびに、プログラム | |
US9619864B2 (en) | Image processing apparatus and method for increasing sharpness of images | |
KR101105675B1 (ko) | 영상 데이터 인페인팅 방법 및 장치 | |
JP6155349B2 (ja) | デコンボリューション画像において色収差を減じる方法、装置及びコンピュータプログラム製品 | |
US20150221100A1 (en) | Information processing apparatus and computer-readable storage medium storing program | |
JP6835227B2 (ja) | 画像処理装置、画像処理方法およびコンピュータプログラム | |
JP6818585B2 (ja) | 画像処理装置、画像処理方法、及び、プログラム | |
JP4966167B2 (ja) | 画像評価装置及びこの画像評価装置を有するカメラ | |
JP5085589B2 (ja) | 画像処理装置および方法 | |
JPWO2013001599A1 (ja) | 商品画像処理装置、商品画像処理方法、情報記録媒体、ならびに、プログラム | |
JP5862370B2 (ja) | 画像処理装置、画像処理方法及びプログラム | |
US11232542B2 (en) | Automated noise attenuation in frequency domains | |
Ghimpeteanu et al. | Three Approaches to Improve Denoising Results that Do Not Involve Developing New Denoising Methods | |
Mehrish et al. | Comprehensive Analysis And Efficiency Comparison Of Image Stitching Techniques |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
ENP | Entry into the national phase |
Ref document number: 2013509350 Country of ref document: JP Kind code of ref document: A |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 12875596 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14356039 Country of ref document: US Ref document number: 2012875596 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |