CN107330885A - A kind of multi-operator image reorientation method of holding important content region the ratio of width to height - Google Patents
A kind of multi-operator image reorientation method of holding important content region the ratio of width to height Download PDFInfo
- Publication number
- CN107330885A CN107330885A CN201710549826.7A CN201710549826A CN107330885A CN 107330885 A CN107330885 A CN 107330885A CN 201710549826 A CN201710549826 A CN 201710549826A CN 107330885 A CN107330885 A CN 107330885A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- image
- mtd
- size
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4007—Interpolation-based scaling, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4023—Decimation- or insertion-based scaling, e.g. pixel or line decimation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06T2207/10004—Still image; Photographic image
Abstract
The invention discloses a kind of multi-operator image reorientation method of holding important content region the ratio of width to height, vision significance detection is carried out to input picture first, and the normalized gray-scale map of combination calculates and obtains energy diagram spectrum, carry out Threshold segmentation to energy collection of illustrative plates and divide to obtain the notable object of vision, in order to reduce the overall dimensions of the notable object of vision, size according to the notable object of vision calculates the stretched dimension for needing to do, and different degrees of stretching is done to the notable object of vision and background information respectively, then similarity transformation is done to the image after stretching, it is finely adjusted finally according to target size using the line clipping method for adding GVF, obtain final output image.By the inventive method, it can detect and the global notable content of protection vision, so as to improve the quality of image redirecting technique.
Description
Technical field
The invention belongs to image processing field, more particularly, to a kind of image reorientation method based on perception of content,
The multi-operator image for keeping important content region the ratio of width to height more particularly to based on perception of content redirects algorithm.
Background technology
Image reorientation method based on perception of content --- line clipping method is for example closest with traditional reorientation method
Interpolation, bilinear interpolation and cutting etc. are compared, and can arbitrarily change the aspect ratio of image without allowing content to become distortion, very
Suitable for the increasingly diversified mobile terminal device of present display size.Existing a few classes are simply introduced below based on line clipping
Image reorientation method.
First kind method, individually carries out image redirection, its reorientation method is in image energy using line clipping technology
Optimization is searched on figure and cuts gap, then cuts gap to change image size by insertion or removal.But this method
Have the disadvantage:Only rely on deletion and cut gap change image size, picture material easily produces zigzag distortion, and then influences vision
The integrality of content.
Equations of The Second Kind method, image redirections, this kind of method are carried out using multiple operations such as line clipping, scaling and tradition cuttings
In there are two methods.First method is simple with reference to line clipping and proportional zoom, passes through the line clipping and ratio of varying number
Example scaling carries out image redirection, specifically utilizes the two-way similarity function of image Euclidean distance, prevailing color
Description similitude and heat input change to determine the optimal number of line clipping.For second regular combination line clipping with than
Example scaling, the loss of content in image redirection process is reduced by a certain degree of reverse utilization line clipping.Specifically cut out
Quantity is confirmed still through using the two-way similarity function of image Euclidean distance.The performance of both approaches be better than only with
The performance of line clipping technology, but their defect is:Visual salient region in image is not detected and protected, easily
Cause the loss of important information, so as to cause the distortion of image.
To sum up, existing image reorientation method fails to embody high-quality visual effect, and the performance of existing algorithm needs
Lifting.
The content of the invention
The shortcoming of high-quality visual effect is embodied it is an object of the invention to overcome conventional images reorientation method to fail,
There is provided a kind of multi-operator image reorientation method of holding important content region the ratio of width to height, higher-quality vision is resulted in
Experience effect, so as to improve the performance of image redirecting technique.
To achieve these goals, the invention provides a kind of multi-operator image weight of holding important content region the ratio of width to height
Orientation method, including:
(1) significance detection is carried out to original image and obtains saliency map, after the saliency map and normalization
Gray-scale map generation energy collection of illustrative plates;
(2) binarization segmentation is carried out to the energy collection of illustrative plates and obtains important content region and background area;
(3) according to the size of target image, and important content region and background area size, calculate it is important in
Hold the size that region and background area should be stretched respectively in non-redirection dimension;
(4) size that should be stretched respectively in non-redirection dimension according to important content region and background area, using uniform
Amplification method does the stretching of non-redirection dimension to important content area, and image background regions are done using reverse line cut-out method
The non-stretching for redirecting dimension, and the stretching for redirecting dimension is done to the image after stretching using uniform amplification method;Wherein, weight
Orientation dimension refers to the dimension of needs scaling, and non-redirection dimension refers to another dimension;
(5) image after the stretching obtained to step (4) is similarity transformation, and image uniform is zoomed into non-redirect ties up
Spend size identical with original image;
(6) image scaling after the scaling that will be obtained in step (5) using the reverse line cut-out method for adding GVF
To target size.
In one embodiment of the present of invention, the step (3) is specially:
(3.1) size delta of the original image in non-redirection dimension is calculated:
Wherein, the h represents size of the original image in non-redirection dimension, the w1Important content region is represented in weight
Orient the size of dimension, the wfRepresent target image and redirecting the size of dimension, and meet wf< w1, the Δ h represents original
Size delta of the beginning image in non-redirection dimension;
(3.2) according to the important content region of the original image increment size different from the pro rate of background area, wherein,
Important content region stretching increment be:
Wherein, the h1Size of the important content region in non-redirection dimension is represented, the β represents auto-adaptive parameter,
The Δ h1Represent the size delta of important content region stretching;
Upper background area and lower background area stretching size delta be respectively:
Wherein, the h2Background area size in representative, the h3Represent lower background area size, the Δ h2In representative
The size delta of background area stretching, the Δ h3Represent the size delta of lower background area stretching.
In one embodiment of the present of invention, background information section is entered using reverse line cut-out method in the step (4)
Row size is stretched, and is specially:
(4.1.1) energizes value to each pixel of background information section image;The Grad for defining image is used as energy
Value:
Wherein, x represents original image,Represent convolution, the gradient map of e representative images.
(4.1.2) finds the minimum cutting wires of energy value with dynamic programming method;
(4.1.3) replicates each pixel of the cutting wires and is inserted into below preimage vegetarian refreshments.
In one embodiment of the present of invention, the dynamic programming method is:
(4.1.2.1) calculates the corresponding forward direction cumulative energy value figure of each pixel in image, and calculating process is as follows:
M (i, j)=e (i, j)+min (M (i-1, j-1), M (i-1, j), M (i-1, j+1))
Wherein, the e (i, j) represents the energy of the pixel, and the M (i, j) represents the forward direction of the pixel calculated
Cumlative energy;
(4.1.2.2) finds out the location point (i, j) of minimum energy value from forward direction cumlative energy figure M last columns, and with
This point is entrance, takes that minimum point of cumlative energy in adjacent in lastrow 3 points to exist as this line clipping route every time
The point of the row, i.e.,:
Seam (i-1)=min (i-1, j-1), (i-1, j), (i-1, j+1) }
Wherein, the seam (i-1) represents the selected point of cutting wires (i-1) row;
(4.1.2.3) is repeated up to search for the first row by this rule can obtain cutting wires.
In one embodiment of the present of invention, the method uniformly amplified in the step (4) is bilinearity differential technique, bilinearity
Interpolation method is calculated using the gray value of image as energy value, and bilinearity differential technique image uniform scaling step is as follows:
(4.2.1) calculates image slices vegetarian refreshments and original image pixel point correspondence after scaling according to scaling target size:
Wherein, the if, jfThe transverse and longitudinal coordinate of image after scaling, the k are represented respectivelyi, kjThe contracting of transverse and longitudinal is represented respectively
Ratio is put, i, j represents the transverse and longitudinal coordinate of original image respectively;
(4.2.2) then pixel (if, jf) corresponding to four neighborhood point energy values be respectively:
Wherein, it is describedWithI is represented respectivelyfAnd jfIt is upper whole;
The calculating of (4.2.3) bilinearity difference:
Wherein, the ef(i, j) represents the pixel energy value in image after scaling;
(4.2.4) calculates the energy value of each pixel of image after scaling, you can obtain zoomed image.
In one embodiment of the present of invention, the step (6) is specially:
(6.1) the GVF vector GVF (i, j) of each pixel in image are calculated, GVF (i, j) includes size and Orientation;
(6.2) with image the first row pixel (1, j) be initial point, downwards line by line planning drawing n bar linear slits, n is image
Redirecting the size of dimension;
(6.3) calculate n bar linear slits accumulated energies value, here can measure GVF vector size;
(6.4) the minimum linear slit of energy value is deleted, and repeat step (6.1) is to (6.4), until zooming to target image
Size, completes compression of images.
In one embodiment of the present of invention, the step (6.2) is specially:
GVF vector directions planning according to each pixel in image draws candidate's cutting wires, it is assumed that current pixel point is
(i, j), then planning downwards, the pixel of (i+1) row pixel (i+1, j-1), (i+1, j), in (i+1, j+1) chooses,
Selection rule is as follows:
Wherein, the θ [GVF (i, j)] represents the direction of the GVF vectors of point (i, j).
In one embodiment of the present of invention, the auto-adaptive parameter β is according to important content region and background area portion
Ratio is chosen:
In one embodiment of the present of invention, entered in the step (4) using uniform amplification method to redirecting dimension direction
Row stretching wants the size of content area to stretch with balance weight, and the size after it is stretched is:
Wherein, the w represents the original size of original image, the wenlRepresent the picture size after stretching.
In one embodiment of the present of invention, the energy collection of illustrative plates is calculated by following formula:
E=S × Gm
Wherein described S represents the saliency map of original image, the GmThe gray-scale map after normalization is represented, wherein:
Wherein, the G represents the gray value of original image, and its span is (0,255), the GmaxAnd GminRespectively
The maximum and minimum value of image intensity value.
In one embodiment of the present of invention, the step (2) is specially:
(2.1) adaptive threshold is chosen, binarization segmentation is carried out to energy collection of illustrative plates according to the adaptive threshold, two are obtained
Value energy collection of illustrative plates;Including:
(2.1.1) chooses adaptive threshold, and the binary-state threshold is so that the maximized T of object function0Value, energy is big
In T0Point be referred to as foreground point, energy is less than T0Point be referred to as background dot, object function is:
G(T0)=ω0(U0-U)2+ω1(U1-U)2
Wherein, the ω0、ω1The number for representing foreground point and background dot respectively accounts for the ratio that whole collection of illustrative plates is always counted, institute
State U0、U1The average energy of foreground point and background dot is represented respectively, and the U represents the average value of view picture collection of illustrative plates energy;
(2.1.2) carries out binarization segmentation according to the adaptive threshold, and partition principle is:
Wherein, the E (x, y) and Em(x, y) represents the energy collection of illustrative plates of input and the binaryzation energy collection of illustrative plates of output respectively;
(2.2) point that detection binaryzation energy collection of illustrative plates intermediate value is 1, chooses the top (i1, j1), bottom (i2, j2), most
Left (i3, j3) and rightmost (i4, j4) four boundary points;
(2.3) then quadrangle [(j1-j2)×(i4-i3)] it is important content region, remaining is then background information section.
The beneficial effects of the invention are as follows:Conventional images method for relocating can be overcome to fail to embody high-quality visual effect
Shortcoming, reached by adjusting the global proportionality of important object in redirection process protect important object purpose, improve figure
As the quality redirected, image redirecting technique is set to obtain higher-quality visual experience effect.
Brief description of the drawings
Fig. 1 is a kind of multi-operator image reorientation method of holding important content region the ratio of width to height in the embodiment of the present invention
Schematic flow sheet;
Fig. 2 is a kind of multi-operator image redirection process of holding important content region the ratio of width to height in the embodiment of the present invention
Image change schematic diagram;
Fig. 3 is that the schematic flow sheet of method is redirected to picture traverse direction in the embodiment of the present invention;
Fig. 4 is that original resolution is 400 × 344, and target resolution compares signal for 280 × 344 image scaling effect
Figure;
Fig. 5 is that original resolution is 267 × 400, and target resolution compares signal for 134 × 400 image scaling effect
Figure;
Fig. 6 is that original resolution is 400 × 300, and target resolution compares signal for 200 × 300 image scaling effect
Figure;
Fig. 7 is that original resolution is 300 × 400, and target resolution compares signal for 210 × 300 image scaling effect
Figure.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.As long as in addition, technical characteristic involved in each embodiment of invention described below
Not constituting conflict each other can just be mutually combined.
As depicted in figs. 1 and 2, reset the invention provides a kind of multi-operator image for keeping important content region the ratio of width to height
To method, including:
(1) significance detection is carried out to original image and obtains saliency map, after the saliency map and normalization
Gray-scale map generation energy collection of illustrative plates;
Specifically, the energy collection of illustrative plates can be calculated by following formula:
E=S × Gm
Wherein, the S represents the saliency map of original image, the GmThe gray-scale map after normalization is represented, wherein:
Wherein, the G represents the gray value of original image, and its value is (0~255), the GmaxAnd GminRespectively image
The maximum and minimum value of gray value.
(2) binarization segmentation is carried out to the energy collection of illustrative plates and obtains important content region and background area;Specifically include:
(2.1) adaptive threshold is chosen, binarization segmentation is carried out to energy collection of illustrative plates according to the adaptive threshold, two are obtained
Value energy collection of illustrative plates;Including:
(2.1.1) chooses adaptive threshold, and the binary-state threshold is so that the maximized T of object function0Value, energy is big
In T0Point be referred to as foreground point, energy is less than T0Point be referred to as background dot.Object function can be expressed as:
G(T0)=ω0(U0-U)2+ω1(U1-U)2
Wherein, the ω0、ω1The number for representing foreground point and background dot respectively accounts for the ratio that whole collection of illustrative plates is always counted, institute
State U0、U1The average energy of foreground point and background dot is represented respectively, and the U represents the average value of view picture collection of illustrative plates energy;
(2.1.2) carries out binarization segmentation according to the adaptive threshold, and partition principle can be expressed as:
Wherein, the E (x, y) and Em(x, y) represents the energy collection of illustrative plates of input and the binaryzation energy collection of illustrative plates of output respectively;
(2.2) point that detection binaryzation energy collection of illustrative plates intermediate value is 1, chooses the top (i1, j1), bottom (i2, j2), most
Left (i3, j3) and rightmost (i4, j4) four boundary points;
(2.3) then quadrangle [(j1-j2)×(i4-i3)] it is important content region, remaining is then background information section;
(3) according to the size of target image, and important content region and background area size, calculate it is important in
Hold the size that region and background area should be stretched respectively in non-redirection dimension;
By the ratio shared by the stretching reduction important content of non-redirection dimension so as to the protection weight in redirection process
The ratio of width to height of content is wanted, for convenience, in the present invention by taking reduced width as an example, its stretching step is as follows:
(3.1) the non-increment for redirecting dimension (longitudinal direction height) is calculated, can be expressed as:
Wherein, the h represents the height of original image, the w1Represent the width in important content region, the wfRepresent mesh
The width of logo image, the Δ h represents longitudinal increment of altitude;
(3.2) according to the important content region of the original image height gain value different from the pro rate of background area,
Wherein, the height gain of important content region stretching is:
Wherein, the h1The height in important content region is represented, the β represents auto-adaptive parameter, the Δ h1Represent important
The height gain of content area stretching;Normally, the auto-adaptive parameter β is according to important content region and background area portion
Ratio is chosen, and can be expressed as:
Upper background area and lower background area stretching height gain be respectively:
Wherein, the h2Background area height, the h in representative3Represent lower background area height, the Δ h2In representative
The height gain of background area stretching, the Δ h3Represent the height gain of lower background area stretching;
(4) size that should be stretched respectively in non-redirection dimension according to important content region and background area, using uniform
Amplification method does the stretching of non-redirection dimension to important content area, and image background regions are done using reverse line cut-out method
The non-stretching for redirecting dimension, and the stretching for redirecting dimension is done to the image after stretching using uniform amplification method;Wherein, weight
Orientation dimension refers to the dimension of needs scaling, and non-redirection dimension refers to another dimension;
(4.1) high elongation is carried out to background information section using reverse line cut-out method;Its step is as follows:
(4.1.1) energizes value to each pixel of background information section image;We determine in reverse line tailoring process
The Grad of adopted image is defined as follows as energy value, gradient:
Then it is applied in digital picture
Wherein, x represents original image,Represent convolution, the gradient map of e representative images.
(4.1.2) finds the minimum cutting wires of energy value with dynamic programming method;Specifically, the dynamic programming method
For:
(4.1.2.1) calculates the corresponding forward direction cumulative energy value figure of each pixel in image, and calculating process is as follows:
M (i, j)=e (i, j)+min (M (i-1, j-1), M (i-1, j), M (i-1, j+1))
Wherein, the e (i, j) represents the energy of the pixel, and the M (i, j) represents the forward direction of the pixel calculated
Cumlative energy;
(4.1.2.2) finds out the location point (i, j) of minimum energy value from forward direction cumlative energy figure M last columns, and with
This point is entrance, takes that minimum point of cumlative energy in adjacent in lastrow 3 points to exist as this line clipping route every time
The point of the row, i.e.,:
Seam (i-1)=min (i-1, j-1), (i-1, j), (i-1, j+1) }
Wherein, the seam (i-1) represents the selected point of cutting wires (i-1) row;
(4.1.2.3) is repeated up to search for the first row by this rule can obtain cutting wires.
(4.1.3) replicates each pixel of the cutting wires and is inserted into below preimage vegetarian refreshments.
(4.2) high elongation is carried out to important content area using the method uniformly amplified;
(4.3) uniform amplification method is used to be stretched the high elongation for wanting content area with balance weight to width, it draws
Width after stretching can be expressed as:
Wherein, the w represents the original width of original image, the wenlRepresent the picture traverse after stretching;
Further, the method uniformly amplified in described (4.2) and (4.3) is bilinearity differential technique, bilinear interpolation
Calculated using the gray value of image as energy value, bilinearity differential technique image uniform scaling step is as follows:
(4.2.1) calculates image slices vegetarian refreshments and original image pixel point correspondence after scaling according to scaling target size:
Wherein, the if, jfThe transverse and longitudinal coordinate of image after scaling, the k are represented respectivelyi, kjThe contracting of transverse and longitudinal is represented respectively
Ratio is put, i, j represents the transverse and longitudinal coordinate of original image respectively;
(4.2.2) then pixel (if, jf) corresponding to four neighborhood point energy values be respectively:
Wherein, it is describedWithI is represented respectivelyfAnd jfIt is upper whole;
The calculating of (4.2.3) bilinearity difference:
Wherein, the ef(i, j) represents the pixel energy value in image after scaling;
(4.2.4) calculates the energy value of each pixel of image after scaling, you can obtain zoomed image.
(5) image after the stretching obtained to step (4) is similarity transformation, and image uniform is zoomed into non-redirect ties up
Spend size identical with original image;
(6) image scaling after the scaling that will be obtained in step (5) using the reverse line cut-out method for adding GVF
To target size, specifically:
(6.1) the GVF vector GVF (i, j) of each pixel in image are calculated, GVF (i, j) includes size and Orientation;
(6.2) with image the first row pixel (1, j) be initial point, downwards line by line planning drawing n bar linear slits, n is image
Width;
Specifically, the GVF vector directions planning according to each pixel in image draws candidate's cutting wires, it is assumed that current picture
Vegetarian refreshments is (i, j), then planning downwards, and the pixel of (i+1) row is in pixel (i+1, j-1), (i+1, j), in (i+1, j+1)
Choose, selection rule is as follows:
Wherein, the θ [GVF (i, j)] represents the direction of the GVF vectors of point (i, j).
(6.3) calculate n bar linear slits accumulated energies value, here can measure GVF vector size;
(6.4) the minimum linear slit of energy value is deleted, and repeat step (6.1) is to (6.4), until zooming to target image
Size, completes compression of images.
Illustrate inventive method with reference to an instantiation, in the following embodiments, only with piece image reduced width
Exemplified by, the multi-operator image reorientation method of holding important content region the ratio of width to height of the present invention is described.Such as Fig. 3 institutes
Show, this method comprises the following steps:
Step 100:Read in original image.
Step 101:The detection of image vision significance is carried out to input picture and obtains saliency map.
Step 102:Gray-scale map is obtained to the greyscale transformation that input picture is normalized.
Step 103:Saliency map is multiplied with gray-scale map correspondence and obtains the energy collection of illustrative plates of input picture.
Step 104:Maximum variance between clusters (OTSU) Threshold segmentation is carried out to energy collection of illustrative plates and obtains important content:
(1041) binary-state threshold is chosen automatically, and threshold value is so that the maximized T of object function0Value, energy is more than T0Point
Referred to as foreground point, energy is less than T0Point be referred to as background dot.Object function can be expressed as:
G(T0)=ω0(UO-U)2+ω1(U1-U)2
Wherein, the ω0、ω1The number for representing foreground point and background dot respectively accounts for the ratio that whole collection of illustrative plates is always counted, institute
State U0、U1The average energy of foreground point and background dot is represented respectively, and the U represents the average value of view picture collection of illustrative plates energy.
(1042) binarization segmentation is carried out according to threshold value, partition principle can be expressed as:
Wherein, the E (x, y) and Em(x, y) represents the energy collection of illustrative plates of input and the binaryzation energy collection of illustrative plates of output respectively.
(1043) point that detection binary image intermediate value is 1, chooses the top (i1, j1), bottom (i2, j2), leftmost
(i3, j3) and rightmost (i4, j4) four boundary points, then quadrangle [(j1-j2)×(i4-i3)] it is important content region.
Step 105:Calculated according to the size, original image size and target image size of the important content of division in important
Hold the level of stretch of region and background area.
(1051) longitudinal increment of altitude, can be expressed as:
Wherein, the height of the h representative images, the width in the w1 representative images important content region, the wf is represented
The target width of image, the Δ h represents longitudinal increment of altitude;
(1052) according to the image important content region increment size different from the pro rate of background information section, wherein,
Important content region stretching height gain be:
Wherein, the h1The height of representative image important content, the β represents auto-adaptive parameter, the Δ h1Represent important
The height gain of content stretching.
Upper background and the height gain of lower background area stretching are respectively:
Wherein, the h2Background area height, the h in representative3Represent lower background area height, the Δ h2In representative
The height gain of background area stretching, the Δ h3Represent the height gain of lower background area stretching;
(1053) high elongation that width tension balance weight wants content area is carried out, the width after it is stretched can be represented
For:
Wherein, the original width of the w representative images, the wenlRepresent the picture traverse after stretching;
(1054) according to the adaptive Selecting All Parameters β value in important content region and background area ratio, it can be expressed as:
Step 106:The size calculated according to step 105, is uniformly amplified to important content area, and background is believed
Breath part increases width using reverse line clipping increase height, using uniform amplification.
Step 107:Image after amplification is carried out similarity transformation to the height such as original image.
Step 108:Image is finely adjusted to target chi using the improved line clipping algorithm that with the addition of GVF
It is very little.
(1081) the GVF vector GVF (i, j) of each pixel in image are calculated, size and Orientation is included;
(1082) with image the first row pixel (1, j) be initial point, downwards line by line planning drawing n bar linear slits, n is image
Width, line by line Plan Rule be:
Assuming that current pixel point be (i, j), then downwards planning, (i+1) row pixel point (i+1, j-1), (i+1,
J), chosen in (i+1, j+1), selection rule is:
Wherein, the θ [GVF (i, j)] told represents the direction of the GVF vectors of point (i, j).
(1083) calculate n bar linear slits accumulated energies value, here can measure GVF vector size;
(1084) delete the minimum linear slit of energy value and repeat step (1081) is to (1084), until zooming to target image
Size, complete compression of images.
Step 109:Export target image.
So far whole adapting to image redirection process is completed, by performing the process, conventional images weight can be overcome
Orientation method fails to embody the shortcoming of high-quality visual effect, and redirection is reached by adjusting the global proportionality of important object
During protect important object purpose so that improve image redirection quality, make image redirecting technique obtain it is more high-quality
The visual experience effect of amount.
In order to test the performance of image reorientation method of the invention, using the common image redirected specifically designed for image
Database carries out experiment test.For the performance of measure algorithm, in terms of the objective scaling quality two of image subjective effect and image
Method and other four kinds of images respectively to the present invention redirects algorithm and has carried out the comparison of performance.
As shown in FIG. 4,5,6, 7, it is the scaling for the four width images that original resolution and target resolution are differed respectively
Results contrast.There is the different distortion of degree in image after being scaled in Fig. 4~Fig. 7 using other method, and distortion content exists
Marked out in each figure with red boxes come.This be primarily due to other method be only single downscaled images width or
Highly, when the notable object of the vision included in image is excessive, the notable object of vision will certainly be influenceed during scaling and is produced
Raw distortion.And the inventive method is changed due to being the notable object of protection vision of overall importance by the stretching to width and height
The overall dimensions of the notable object of vision adapt it to target size, so that vision notable object is intactly preserved, therefore just not
The distortion phenomenon of other method can be produced.Above-mentioned analysis shows, using the subjective effect of the inventive method reconstruction image than other
Method is more preferable.
Objective evaluation is carried out to the performance of the inventive method using image scaling criteria of quality evaluation, the evaluation criterion passes through
Travel through the performance figure that the correlation of original image and zoomed image in different scale space obtains image quality evaluation, the quality
The scope of index is [0,1], and the matching degree of the bigger explanation two images of numerical value is higher, i.e. the quality of zoomed image is better.Analysis
The picture quality index for comparing 15 width associated pictures is as shown in table 1 below.
The picture quality index of table 1 compares
SC | GVF | MO | OUR | |
Fig. 2 | 0.0970 | 0.1358 | 0.1583 | 0.1594 |
Fig. 3 | 0.4029 | 0.3602 | 0.3840 | 0.4469 |
Fig. 4 | 0.3124 | 0.4418 | 0.2694 | 0.5478 |
Fig. 5 | 0.1498 | 0.1783 | 0.0354 | 0.1858 |
The above results show that the picture quality after being scaled using the inventive method is substantially better than what is obtained using other method
Image, this strengthens the protection of the notable content of vision mainly due to the inventive method, and ensure that the complete of the notable content of vision
Property so that correlation of the image with original image spatially after scaling is very high, and matching degree is also significantly improved therewith.Thus
Illustrate, the inventive method equally has better performance in objective evaluation result.
As it will be easily appreciated by one skilled in the art that the foregoing is merely illustrative of the preferred embodiments of the present invention, it is not used to
The limitation present invention, any modifications, equivalent substitutions and improvements made within the spirit and principles of the invention etc., it all should include
Within protection scope of the present invention.
Claims (10)
1. the multi-operator image reorientation method of a kind of holding important content region the ratio of width to height, it is characterised in that including following step
Suddenly:
(1) significance detection is carried out to original image and obtains saliency map, utilize the ash after the saliency map and normalization
Degree figure generation energy collection of illustrative plates;
(2) binarization segmentation is carried out to the energy collection of illustrative plates and obtains important content region and background area;
(3) according to the size of target image, and important content region and background area size, calculate important content area
The size that domain and background area should be stretched respectively in non-redirection dimension;
(4) size that should be stretched respectively in non-redirection dimension according to important content region and background area, using uniform amplification
Method does the stretching of non-redirection dimension to important content area, and image background regions are done with non-heavy using reverse line cut-out method
The stretching of dimension is oriented, and does the stretching for redirecting dimension to the image after stretching using uniform amplification method;Wherein, redirect
Dimension refers to the dimension of needs scaling, and non-redirection dimension refers to another dimension;
(5) image after the stretching obtained to step (4) is similarity transformation, and it is big that image uniform is zoomed into non-redirection dimension
It is small identical with original image;
(6) image scaling is to target chi after the scaling that will be obtained in step (5) using the line clipping method for adding GVF
It is very little.
2. the multi-operator image reorientation method of holding important content region according to claim 1 the ratio of width to height, its feature
It is, the step (3) is specially:
(3.1) size delta of the original image in non-redirection dimension is calculated:
<mrow>
<mi>&Delta;</mi>
<mi>h</mi>
<mo>=</mo>
<mi>h</mi>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mfrac>
<msub>
<mi>w</mi>
<mn>1</mn>
</msub>
<msub>
<mi>w</mi>
<mi>f</mi>
</msub>
</mfrac>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
Wherein, the h represents size of the original image in non-redirection dimension, the w1Important content region is represented in redirection
The size of dimension, the wfRepresent target image and redirecting the size of dimension, and meet wf< w1, the Δ h represents original graph
As the size delta in non-redirection dimension;
(3.2) according to the important content region of the original image increment size different from the pro rate of background area, wherein, it is important
Content area stretching increment be:
<mrow>
<msub>
<mi>&Delta;h</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<mi>&beta;</mi>
<mo>&times;</mo>
<mfrac>
<msub>
<mi>h</mi>
<mn>1</mn>
</msub>
<mi>h</mi>
</mfrac>
<mo>&times;</mo>
<mi>&Delta;</mi>
<mi>h</mi>
</mrow>
Wherein, the h1Size of the important content region in non-redirection dimension is represented, the β represents auto-adaptive parameter, the Δ
h1Represent the size delta of important content region stretching;
Upper background area and lower background area stretching size delta be respectively:
<mrow>
<msub>
<mi>&Delta;h</mi>
<mn>2</mn>
</msub>
<mo>=</mo>
<mi>&Delta;</mi>
<mi>h</mi>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>&beta;</mi>
<mo>&times;</mo>
<mfrac>
<msub>
<mi>h</mi>
<mn>1</mn>
</msub>
<mi>h</mi>
</mfrac>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mfrac>
<msub>
<mi>h</mi>
<mn>2</mn>
</msub>
<mrow>
<msub>
<mi>h</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<msub>
<mi>h</mi>
<mn>3</mn>
</msub>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>&Delta;h</mi>
<mn>3</mn>
</msub>
<mo>=</mo>
<mi>&Delta;</mi>
<mi>h</mi>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>&beta;</mi>
<mo>&times;</mo>
<mfrac>
<msub>
<mi>h</mi>
<mn>1</mn>
</msub>
<mi>h</mi>
</mfrac>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mfrac>
<msub>
<mi>h</mi>
<mn>3</mn>
</msub>
<mrow>
<msub>
<mi>h</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<msub>
<mi>h</mi>
<mn>3</mn>
</msub>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
Wherein, the h2Background area size in representative, the h3Represent lower background area size, the Δ h2Background in representative
The size delta of region stretching, the Δ h3Represent the size delta of lower background area stretching;
3. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, it is special
Levy and be, size stretching is carried out to background information section using reverse line cut-out method in the step (4), is specially:
(4.1.1) energizes value to each pixel of background information section image;The Grad for defining image is used as energy value:
<mrow>
<mi>e</mi>
<mo>=</mo>
<mi>x</mi>
<mo>&CircleTimes;</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>2</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>2</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>+</mo>
<mi>x</mi>
<mo>&CircleTimes;</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>2</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>2</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, x represents original image,Represent convolution, the gradient map of e representative images.
(4.1.2) finds the minimum cutting wires of energy value with dynamic programming method;
(4.1.3) replicates each pixel of the cutting wires and is inserted into below preimage vegetarian refreshments.
4. the multi-operator image reorientation method of holding important content region according to claim 3 the ratio of width to height, its feature
It is, the dynamic programming method is:
(4.1.2.1) calculates the corresponding forward direction cumulative energy value figure of each pixel in image, and calculating process is as follows:
M (i, j)=e (i, j)+min (M (i-1, j-1), M (i-1, j), M (i-1, j+1))
Wherein, the e (i, j) represents the energy of the pixel, and the M (i, j) represents the forward direction accumulation of the pixel calculated
Energy;
(4.1.2.2) finds out the location point (i, j) of minimum energy value from forward direction cumlative energy figure M last columns, and with this point
For entrance, that minimum point of cumlative energy in adjacent in lastrow 3 points is taken to be used as this line clipping route in the row every time
Point, i.e.,:
Seam (i-1)=min (i-1, j-1), (i-1, j), (i-1, j+1) }
Wherein, the seam (i-1) represents cutting wires (i-1 (the selected points of row;
(4.1.2.3) is repeated up to search for the first row by this rule can obtain cutting wires.
5. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, it is special
Levy and be, the method uniformly amplified in the step (4) is bilinearity differential technique, bilinear interpolation utilizes the gray value of image
Calculated as energy value, bilinearity differential technique image uniform scaling step is as follows:
(4.2.1) calculates image slices vegetarian refreshments and original image pixel point correspondence after scaling according to scaling target size:
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<msub>
<mi>i</mi>
<mi>f</mi>
</msub>
<mo>=</mo>
<mi>i</mi>
<mo>/</mo>
<msub>
<mi>k</mi>
<mi>i</mi>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>j</mi>
<mi>f</mi>
</msub>
<mo>=</mo>
<mi>j</mi>
<mo>/</mo>
<msub>
<mi>k</mi>
<mi>j</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein, the if, jfThe transverse and longitudinal coordinate of image after scaling, the k are represented respectivelyi, kjThe pantograph ratio of transverse and longitudinal is represented respectively
Example, i, j represents the transverse and longitudinal coordinate of original image respectively;
(4.2.2) then pixel (if, jf) corresponding to four neighborhood point energy values be respectively:
Wherein, it is describedWithI is represented respectivelyfAnd jfIt is upper whole;
The calculating of (4.2.3) bilinearity difference:
Wherein, the ef(i, j) represents the pixel energy value in image after scaling;
(4.2.4) calculates the energy value of each pixel of image after scaling, you can obtain zoomed image.
6. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, it is special
Levy and be, the step (6) is specially:
(6.1) the GVF vector GVF (i, j) of each pixel in image are calculated, GVF (i, j) includes size and Orientation;
(6.2) with image the first row pixel (1, j) be initial point, planning drawing n bar linear slits line by line downwards, n is image in weight
Orient the size of dimension;
(6.3) calculate n bar linear slits accumulated energies value, here can measure GVF vector size;
(6.4) the minimum linear slit of energy value, and repeat step (6.1) to (6.4), the chi until zooming to target image are deleted
It is very little, complete compression of images.
7. the multi-operator image reorientation method of holding important content region according to claim 6 the ratio of width to height, its feature
It is, the step (6.2) is specially:
GVF vector directions planning according to each pixel in image draws candidate's cutting wires, it is assumed that current pixel point for (i,
J), then planning downwards, the pixel of (i+1) row is in pixel (i+1, j-1), (i+1 j), in (i+1, j+1) chooses, chosen
Rule is as follows:
Wherein, the θ [GVF (i, j)] represents the direction of the GVF vectors of point (i, j).
8. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, it is special
Levy and be, use uniform amplification method to stretch redirection dimension direction in the step (4) and content regions are wanted with balance weight
The size in domain is stretched, and the size after it is stretched is:
<mrow>
<msub>
<mi>w</mi>
<mrow>
<mi>e</mi>
<mi>n</mi>
<mi>l</mi>
</mrow>
</msub>
<mo>=</mo>
<mi>w</mi>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>+</mo>
<mi>&beta;</mi>
<mo>&times;</mo>
<mfrac>
<mrow>
<mi>&Delta;</mi>
<mi>h</mi>
</mrow>
<mi>h</mi>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
Wherein, the w represents the original size of original image, the wenlRepresent the picture size after stretching.
9. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, it is special
Levy and be, the energy collection of illustrative plates is calculated by following formula:
E=S × Gm
Wherein described S represents the saliency map of original image, the GmThe gray-scale map after normalization is represented, wherein:
<mrow>
<msub>
<mi>G</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mi>G</mi>
<mo>-</mo>
<msub>
<mi>G</mi>
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
</mrow>
</msub>
</mrow>
<mrow>
<msub>
<mi>G</mi>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>G</mi>
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
</mrow>
</msub>
</mrow>
</mfrac>
</mrow>
Wherein, the G represents the gray value of original image, and span is (0,255), the GmaxAnd GminRespectively image is grey
The maximum and minimum value of angle value.
10. the multi-operator image reorientation method of holding important content region according to claim 1 or 2 the ratio of width to height, its
It is characterised by, the step (2) is specially:
(2.1) adaptive threshold is chosen, binarization segmentation is carried out to energy collection of illustrative plates according to the adaptive threshold, binaryzation is obtained
Energy collection of illustrative plates;Including:
(2.1.1) chooses adaptive threshold, and the binary-state threshold is so that the maximized T of object function0Value, energy is more than T0
Point be referred to as foreground point, energy is less than T0Point be referred to as background dot, object function is:
G(T0)=ω0(U0-U)2+ω1(U1-U)2
Wherein, the ω0、ω1The number for representing foreground point and background dot respectively accounts for the ratio that whole collection of illustrative plates is always counted, the U0、
U1The average energy of foreground point and background dot is represented respectively, and the U represents the average value of view picture collection of illustrative plates energy;
(2.1.2) carries out binarization segmentation according to the adaptive threshold, and partition principle is:
<mrow>
<msub>
<mi>E</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mrow>
<mi>E</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&GreaterEqual;</mo>
<msub>
<mi>T</mi>
<mi>O</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mi>E</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo><</mo>
<msub>
<mi>T</mi>
<mi>O</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, the E (x, y) and Em(x, y) represents the energy collection of illustrative plates of input and the binaryzation energy collection of illustrative plates of output respectively;
(2.2) point that detection binaryzation energy collection of illustrative plates intermediate value is 1, chooses the top (i1, j1), bottom (i2, j2), leftmost
(i3, j3) and rightmost (i4, j4) four boundary points;
(2.3) then quadrangle [(j1-j2)×(i4-i3)] it is important content region, remaining is then background area.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710549826.7A CN107330885B (en) | 2017-07-07 | 2017-07-07 | Multi-operator image redirection method for keeping aspect ratio of important content area |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710549826.7A CN107330885B (en) | 2017-07-07 | 2017-07-07 | Multi-operator image redirection method for keeping aspect ratio of important content area |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107330885A true CN107330885A (en) | 2017-11-07 |
CN107330885B CN107330885B (en) | 2020-10-02 |
Family
ID=60197198
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710549826.7A Active CN107330885B (en) | 2017-07-07 | 2017-07-07 | Multi-operator image redirection method for keeping aspect ratio of important content area |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107330885B (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108665413A (en) * | 2018-04-28 | 2018-10-16 | 中国科学院上海高等研究院 | A kind of image adaptive amplification method, readable storage medium storing program for executing and terminal |
CN109191456A (en) * | 2018-09-19 | 2019-01-11 | 电子科技大学 | Lung CT image processing method and system based on two-dimentional S-transformation |
CN109447970A (en) * | 2018-10-30 | 2019-03-08 | 河北工业大学 | The image reorientation method based on energy transfer and uniformly scaled |
CN109493275A (en) * | 2018-11-16 | 2019-03-19 | 南通大学 | Reorientation method is cut in a kind of fusion notable figure and the seaming and cutting of depth map |
CN109600667A (en) * | 2018-12-05 | 2019-04-09 | 广西大学 | A method of the video based on grid and frame grouping redirects |
CN109615670A (en) * | 2018-12-05 | 2019-04-12 | 广西大学 | A kind of multi-operator image reorientation method based on compression domain perception of content |
CN109712164A (en) * | 2019-01-17 | 2019-05-03 | 上海携程国际旅行社有限公司 | Image intelligent cut-out method, system, equipment and storage medium |
CN109978858A (en) * | 2019-03-27 | 2019-07-05 | 华南理工大学 | A kind of double frame thumbnail image quality evaluating methods based on foreground detection |
CN110580678A (en) * | 2019-09-10 | 2019-12-17 | 北京百度网讯科技有限公司 | image processing method and device |
CN111353382A (en) * | 2020-01-10 | 2020-06-30 | 广西大学 | Intelligent cutting video redirection method based on relative displacement constraint |
CN112529765A (en) * | 2019-09-02 | 2021-03-19 | 阿里巴巴集团控股有限公司 | Image processing method, apparatus and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090219302A1 (en) * | 2008-02-29 | 2009-09-03 | Hong Fu Jin Precision Industry (Shenzhen) Co., Ltd. | Electronic device capable of processing image and method thereof |
WO2013104819A1 (en) * | 2012-01-09 | 2013-07-18 | Nokia Corporation | Method and apparatus for image scaling in photography |
CN103327261A (en) * | 2013-06-04 | 2013-09-25 | 北京工业大学 | Method for simply and fast redirecting video |
US20150052439A1 (en) * | 2013-08-19 | 2015-02-19 | Kodak Alaris Inc. | Context sensitive adaptable user interface |
CN105528591A (en) * | 2016-01-14 | 2016-04-27 | 电子科技大学 | Living fingerprint identification method based on multi-quadrant coding |
US9530184B1 (en) * | 2007-10-26 | 2016-12-27 | Google Inc. | N-patch image resizing |
-
2017
- 2017-07-07 CN CN201710549826.7A patent/CN107330885B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9530184B1 (en) * | 2007-10-26 | 2016-12-27 | Google Inc. | N-patch image resizing |
US20090219302A1 (en) * | 2008-02-29 | 2009-09-03 | Hong Fu Jin Precision Industry (Shenzhen) Co., Ltd. | Electronic device capable of processing image and method thereof |
WO2013104819A1 (en) * | 2012-01-09 | 2013-07-18 | Nokia Corporation | Method and apparatus for image scaling in photography |
CN103327261A (en) * | 2013-06-04 | 2013-09-25 | 北京工业大学 | Method for simply and fast redirecting video |
US20150052439A1 (en) * | 2013-08-19 | 2015-02-19 | Kodak Alaris Inc. | Context sensitive adaptable user interface |
CN105528591A (en) * | 2016-01-14 | 2016-04-27 | 电子科技大学 | Living fingerprint identification method based on multi-quadrant coding |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108665413A (en) * | 2018-04-28 | 2018-10-16 | 中国科学院上海高等研究院 | A kind of image adaptive amplification method, readable storage medium storing program for executing and terminal |
CN109191456A (en) * | 2018-09-19 | 2019-01-11 | 电子科技大学 | Lung CT image processing method and system based on two-dimentional S-transformation |
CN109447970B (en) * | 2018-10-30 | 2020-12-29 | 河北工业大学 | Image redirection method based on energy transfer and uniform scaling |
CN109447970A (en) * | 2018-10-30 | 2019-03-08 | 河北工业大学 | The image reorientation method based on energy transfer and uniformly scaled |
CN109493275A (en) * | 2018-11-16 | 2019-03-19 | 南通大学 | Reorientation method is cut in a kind of fusion notable figure and the seaming and cutting of depth map |
CN109600667A (en) * | 2018-12-05 | 2019-04-09 | 广西大学 | A method of the video based on grid and frame grouping redirects |
CN109615670A (en) * | 2018-12-05 | 2019-04-12 | 广西大学 | A kind of multi-operator image reorientation method based on compression domain perception of content |
CN109600667B (en) * | 2018-12-05 | 2021-04-13 | 广西大学 | Video redirection method based on grid and frame grouping |
CN109712164A (en) * | 2019-01-17 | 2019-05-03 | 上海携程国际旅行社有限公司 | Image intelligent cut-out method, system, equipment and storage medium |
CN109978858B (en) * | 2019-03-27 | 2021-02-19 | 华南理工大学 | Double-frame thumbnail image quality evaluation method based on foreground detection |
CN109978858A (en) * | 2019-03-27 | 2019-07-05 | 华南理工大学 | A kind of double frame thumbnail image quality evaluating methods based on foreground detection |
CN112529765A (en) * | 2019-09-02 | 2021-03-19 | 阿里巴巴集团控股有限公司 | Image processing method, apparatus and storage medium |
CN110580678A (en) * | 2019-09-10 | 2019-12-17 | 北京百度网讯科技有限公司 | image processing method and device |
CN111353382A (en) * | 2020-01-10 | 2020-06-30 | 广西大学 | Intelligent cutting video redirection method based on relative displacement constraint |
CN111353382B (en) * | 2020-01-10 | 2022-11-08 | 广西大学 | Intelligent cutting video redirection method based on relative displacement constraint |
Also Published As
Publication number | Publication date |
---|---|
CN107330885B (en) | 2020-10-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107330885A (en) | A kind of multi-operator image reorientation method of holding important content region the ratio of width to height | |
CN104517110B (en) | The binarization method and system of a kind of image in 2 D code | |
CN112288658A (en) | Underwater image enhancement method based on multi-residual joint learning | |
CN105488758A (en) | Image scaling method based on content awareness | |
CN108537782B (en) | Building image matching and fusing method based on contour extraction | |
CN104835114A (en) | Image self-adaptive display method | |
CN108319894A (en) | Fruit recognition methods based on deep learning and device | |
CN104899845A (en) | Method for fusing multiple exposure images based on 1 alphabeta space scene migration | |
CN110321769A (en) | A kind of more size commodity on shelf detection methods | |
US20030200236A1 (en) | Curve tracing system | |
CN102708538A (en) | Re-sizing image sequences | |
CN104123554B (en) | SIFT image characteristic extracting methods based on MMTD | |
CN104517262B (en) | The adaptive image scaling method detected based on DCT domain vision significance | |
CN106023276B (en) | Pencil drawing method for drafting and device based on image procossing | |
CN104392433B (en) | Image-based digital oil painting canvas automatic generation method | |
CN111507334A (en) | Example segmentation method based on key points | |
CN104992403A (en) | Hybrid operator image redirection method based on visual similarity measurement | |
Garg et al. | Structure preservation of image using an efficient content-aware image retargeting technique | |
CN109345536A (en) | A kind of image superpixel dividing method and its device | |
CN109447970B (en) | Image redirection method based on energy transfer and uniform scaling | |
CN110111342B (en) | Optimized selection method and device for matting algorithm | |
CN104392209B (en) | A kind of image complexity evaluation method of target and background | |
CN111080723A (en) | Image element segmentation method based on Unet network | |
Ružić et al. | Texture and color descriptors as a tool for context-aware patch-based image inpainting | |
CN104268845A (en) | Self-adaptive double local reinforcement method of extreme-value temperature difference short wave infrared image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20211117 Address after: 530004 room 901, floor 9, Qinghu center, No. 36, Zhuxi Avenue, Nanning, Guangxi Zhuang Autonomous Region Patentee after: Guangxi Xinhao Zhiyun Technology Co.,Ltd. Address before: 530004 100 East University Road, XiXiangTang District, Nanning, the Guangxi Zhuang Autonomous Region Patentee before: GUANGXI University |
|
TR01 | Transfer of patent right |