CN102375990A - Method and equipment for processing images - Google Patents
Method and equipment for processing images Download PDFInfo
- Publication number
- CN102375990A CN102375990A CN2010102576281A CN201010257628A CN102375990A CN 102375990 A CN102375990 A CN 102375990A CN 2010102576281 A CN2010102576281 A CN 2010102576281A CN 201010257628 A CN201010257628 A CN 201010257628A CN 102375990 A CN102375990 A CN 102375990A
- Authority
- CN
- China
- Prior art keywords
- image block
- image
- neighbour
- search
- collection
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and equipment for processing images. The method for processing the images comprises the following steps of: searching one or more neighboring image blocks of all the image blocks from an image block set of a plurality of image blocks which are extracted from a sample image, wherein the one or more neighboring image blocks comprise one or more image blocks, which are the closest to all the image blocks, in the image block set; forming a neighboring set of all the image blocks; and then, storing the image block set and information of the neighboring set related to all the image blocks.
Description
Technical field
The disclosure relates to image processing field, in particular to image process method and equipment.
Background technology
In image processing method based on study, usually need obtain the training set (or dictionary) of image or image block through training great amount of images sample, use in order to subsequent image processing.For example; Through training a large amount of low resolution that corresponds to each other and high-definition picture sample; Can obtain to comprise many to the low-resolution image piece that corresponds to each other and the training set of high-definition picture piece, such training set can be used as co-occurrence patterns (Co-occurrence Model) or dictionary and the image super-resolution that is used for utilizing the image of low resolution to generate high-definition picture is used.
Summary of the invention
Provide hereinafter about brief overview of the present disclosure, so that the basic comprehension about some aspect of the present disclosure is provided.Should be appreciated that this general introduction is not about exhaustive general introduction of the present disclosure.It is not that intention is confirmed key of the present disclosure or pith, neither be intended to limit the scope of the present disclosure.Its purpose only is to provide some notion with the form of simplifying, with this as the preorder in greater detail of argumentation after a while.
According to embodiments more of the present disclosure, can in the image block set that comprises a plurality of image blocks that from sample image, extract, search for one or more neighbour's image block of each image block.These one or more neighbour's image blocks comprise one or more nearest image block of said image block set this each image block of middle distance, thereby form neighbour's collection of this each image block.Then, can preserve the information of neighbour's collection of said image block set and relevant each image block.
According to other embodiments of the present disclosure, can in the image block set that comprises a plurality of image blocks that from sample image, extract, carry out coarse search, obtain and the approximate nearest image block of query object distance.Then, can coarse search to the neighbour of image block concentrate and to search one or more image block nearest with said query object, as one or more arest neighbors image block of said query object.Wherein, coarse search to the neighbour collection of image block can comprise one or more nearest image block of said image block set this image block of middle distance.
Description of drawings
With reference to below in conjunction with the explanation of accompanying drawing, can understand above and other purpose, characteristics and advantage of the present disclosure to disclosure embodiment with being more prone to.Parts in the accompanying drawing are just in order to illustrate principle of the present disclosure.In the accompanying drawings, identical or similar techniques characteristic or parts will adopt identical or similar Reference numeral to represent.
Fig. 1 shows the indicative flowchart according to the image processing method of an embodiment of the present disclosure;
Fig. 2 shows the indicative flowchart according to the image processing method of another embodiment of the present disclosure;
Fig. 3 shows the indicative flowchart of the method for neighbour's collection of searching for each image block according to a specific embodiment of the present disclosure, in the image block set;
Fig. 4 shows according to a concrete example of the present disclosure, image block is gathered the synoptic diagram that recurrence is divided into the process of a plurality of subclass;
Fig. 5 shows the synoptic diagram according to the image block set that is organized into tree construction of a concrete example of the present disclosure;
Fig. 6 shows the image block set is divided into the actual divisional plane of two sub-set and the synoptic diagram of desirable divisional plane;
Fig. 7 shows the synoptic diagram of setting up the tabulation of pseudo-point according to a specific embodiment of the present disclosure;
Fig. 8 shows the indicative flowchart according to the image processing method of an embodiment of the present disclosure;
Fig. 9 shows the indicative flowchart according to the image processing method of another embodiment of the present disclosure;
Figure 10 shows the schematic block diagram according to the image processing equipment of an embodiment of the present disclosure;
Figure 11 shows the schematic block diagram according to the image processing equipment of another embodiment of the present disclosure;
Figure 12 shows the schematic block diagram according to the image processing equipment of another embodiment of the present disclosure;
Figure 13 shows 4 width of cloth images; Wherein (a) representes a width of cloth low-resolution image; (d) expression actual photographed, with the corresponding high-definition picture of image (a); (b) expression utilize that bilinear interpolation method generates, with the corresponding high-definition picture of image (a), (c) the expression utilization according to the method for embodiment of the present disclosure or equipment generation, with the corresponding high-definition picture of image (a); And
Figure 14 shows and can be used for implementing the schematic block diagram according to the computing machine of embodiment of the present disclosure.
Embodiment
Embodiment of the present disclosure is described with reference to the accompanying drawings.Element of in an accompanying drawing of the present disclosure or a kind of embodiment, describing and characteristic can combine with element and the characteristic shown in one or more other accompanying drawing or the embodiment.Should be noted that for purpose clearly, omitted and the disclosure parts irrelevant, that those of ordinary skills are known and the expression and the description of processing in accompanying drawing and the explanation.
Fig. 1 shows the indicative flowchart according to the image processing method of an embodiment of the present disclosure.Wherein, Gathering (or being called training set or dictionary) afterwards through training a plurality of sample images to extract image block; Neighbour's image block of each image block in the image block set in this set of search; Thereby generate neighbour's collection of each image block, so that inquiry is used in the subsequent image processing process.
As shown in Figure 1, this method can comprise step 103 and 109.
In step 103, one or more the nearest image block of each image block in the image block set that comprises a plurality of image blocks that from sample image, extract in this set of detection range.The image block that searches is called one or more neighbour's image block of each image block.These neighbour's image blocks form neighbour's collection of each image block.
In the disclosure; Described image block set can be the training set through adopting any method to train the great amount of samples image to obtain; For example; The local energy of each image block in can the analyzing samples image also only keeps and significant image primitive (like edge, turning and/or the spot etc.) image block that corresponding energy is high, does not do to limit and detail at this.
In step 109, can preserve the set of this image block and collect relevant information with the neighbour of each image block in this set.
Utilize method shown in Figure 1, obtain comprising the training set of a plurality of image blocks through training after, can also generate neighbour's collection of each image block in this training set.When preserving this training set, can also preserve the wherein information of neighbour's collection of each image block.Like this; In subsequent image processing, need this training set of inquiry fashionable; After in training set, tentatively searching certain image block, can also be as required and concentrate its neighbour and to carry out search refinement further, thus the image block that is more suited the requirements.Compare with the method that only generates with preserving the training set, method shown in Figure 1 more helps follow-up inquiry, can inquire about the information more accurately that obtains.
Fig. 2 shows the indicative flowchart according to the image processing method of another embodiment of the present disclosure.This embodiment is similar with method shown in Figure 1, and difference is that the image block set can be organized into the tree structure of classification in method shown in Figure 2.
As shown in Figure 2, this method can comprise step 201,203 and 209.Particularly, in step 201, can at first the image block set that comprises a plurality of image blocks that from sample image, extract be organized into multistage tree construction.In this tree construction, root node can the correspondence image set of blocks, and comprises the information that is used for this set is divided into the thresholding of the subclass corresponding with the next stage node.One sub-set of each branch node correspondence image set of blocks and comprising is used for the subclass corresponding with this branch node is divided into the information with the thresholding of the corresponding subclass of next stage node; And one or set of diagrams picture piece in each leaf node correspondence image set of blocks.
As a specific embodiment, can utilize optimum lineoid to come recurrence to cut apart said image block set, thereby form above-mentioned multistage tree construction.Fig. 4 shows a concrete example utilizing optimum lineoid to come recurrence split image set of blocks.For example, can adopt the breadth extreme that uses in the kd tree to cut apart rule (Maximum Spread Splitting Rule) comes recurrence to cut apart said image block set.The image block set of supposing to cut apart is represented with S, then cuts apart the dimension with breadth extreme that dimension can adopt this S set, then can adopt this S set to cut apart the median point of dimension along this at this cut-point of cutting apart on the dimension.Shown in Fig. 4 (A), through divisional plane P401, S set is divided into two types, i.e. class 1 and class 2.Then, shown in Fig. 4 (B), utilize divisional plane P402, class 1 is type of a being divided into 1-1 and a type 1-2 again.Shown in Fig. 4 (C), utilize divisional plane P403, type 1-1 type of a being divided into 1-1-1 and a type 1-1-2.Shown in Fig. 4 (D), utilize divisional plane P404, type 1-1-1 is further type of a being divided into 1-1-1-1 and a type 1-1-1-2 again.The rest may be inferred, and it highly is log that the mode that adopts this recurrence to cut apart, image block S set are divided into
2The binary tree structure of N, the wherein number of the image block among the N presentation video set of blocks S.
Fig. 5 shows an example of the image block set that is organized into multistage tree construction.Wherein, each root node or branch node can comprise the segmentation threshold information that the set of the image block of correspondence or subclass is divided into two types.
As a concrete example, each root node or branch node can comprise following information:
Cut_dim: with the information of the dimension (promptly cutting apart dimension) of divisional plane quadrature;
Cut_val: the position of divisional plane (segmentation threshold value);
Low_bnd: the image block set corresponding or the lower limit of subclass with this tree node;
High_bnd: the image block set corresponding or the upper limit of subclass with this tree node;
Left_child: the link of arriving the left child node of this node;
Right_child: the link of arriving the right child node of this node.
In addition, an element in each leaf node correspondence image set of blocks, i.e. an image block.Each leaf node can comprise following information:
Index: the index of the pairing image block of this leaf node in the image block set.
It should be understood, of course, that above-mentioned tree construction only is exemplary, and be not to be intended to embodiment of the present disclosure is limited to this.Can adopt any other suitable tree constructions, not enumerate one by one here.
In the embodiment shown in Figure 2; Because tree construction has been adopted in the image block set; Therefore; When in the subsequent image processing process, needing this image block set of inquiry, can be at first in tree construction, find leaf node fast with the query object approximate match according to the segmentation threshold information in each tree node.In other words, adopt such tree construction, can carry out traversal search, thereby can improve searching speed greatly whole tree structure.Certainly, owing in the process of recurrence split image set of blocks, cut apart only certain dimension in representative image set of blocks space of the divisional plane that adopted at every turn, the leaf node that therefore searches at this tree construction might not be the image block that matees most with query object.But, can further optimize lookup result through neighbour's collection of the leaf node that further search found.
In some instances, can utilize any appropriate method to search for neighbour's image block of each image block.For example, (as in the step 103) can adopt methods such as k nearest neighbor (K Nearest Neighbors) search procedure to search for neighbour's image block of certain image block, do not enumerate one by one here.Fig. 3 provides the indicative flowchart according to the method for neighbour's image block of in the image block set, searching for certain image block of a specific embodiment of the present disclosure.In the searching method of Fig. 3, one group of pseudo-point is linked with each image block, form neighbour's collection of this each image block.
Suppose that image block P is an image block in the image block set, as shown in Figure 3, the method (like step 103) of neighbour's image block of this image block of search P can comprise step 103-2 to 103-7 in the image block set.In step 103-2, generate a proper vector at random.The dimension of this proper vector is identical with the pixel count that image block P comprises; Be that the two is in same feature space; And the distance of this proper vector and image block P is in predetermined scope, and promptly this proper vector (can be selected one dimension or the apteryx in the image block ensemble space according to actual conditions at certain one dimension or certain apteryx of image block ensemble space; The dimension of selecting is many more, and the neighbour's image block that then utilizes this proper vector to find is accurate more, does not limit here) on and the distance of image block be no more than certain threshold value.Should be understood that this predetermined scope or threshold value can set according to the practical application scene, do not limit here.
In step 103-3, the proper vector that is generated as query object, is searched the image block nearest apart from this query object in said image block set.For example, can adopt any suitable searching method (like exhaustive search algorithm, exhaustive search), not be described further here at this arest neighbors image block of image block set search.
In step 103-4, judge whether the arest neighbors image block that finds is identical with this image block P, if not, then forwards step 104-6 to; If then in step 103-5, this arest neighbors image block as a pseudo-point, and is increased the count value related with this arest neighbors image block; Particularly, if the arest neighbors image block that epicycle searches with before a certain take turns or several arest neighbors image block that obtains in the search of taking turns identical, then its count value is increased one.If the arest neighbors image block that epicycle searches is all inequality with the arest neighbors image block that search before obtains, then its count value is set to one.
In step 103-6, judge whether search has carried out the T wheel, if, then forward step 103-7 to, otherwise, continue to repeat step 103-2 to 103-5, promptly repeat the said step of searching of the said vector sum of generating feature at random.Here T is the integer greater than 1, and concrete value can be set according to actual needs.That is to say; The generating feature vector and the step of in the image block set, searching the arest neighbors image block of this proper vector can be carried out repeatedly repeatedly at random; And can confirm to repeat the number of times of the step that generating feature vector sum at random searches according to the application scenarios of reality, generally speaking, the number of times of repetition is many more; The result who obtains is accurate more, but calculated amount is also big more.Therefore, can weigh the value of T according to actual needs, not limit here.
In step 103-7, with one or more that find, related arest neighbors image block, as neighbour's collection of this image block P with bigger count value.
In method shown in Figure 3; Because the proper vector as query object that generates at random has only one dimension or apteryx, therefore, search for neighbour's image block with the image block that adopts multidimensional as query object and compare; Can significantly reduce calculated amount, thereby improve the speed of search.
As a specific embodiment, searching method shown in Figure 3 can also be used in combination with the method that the image block set is organized into tree construction shown in Figure 2.As stated, might not the image block that matees most with query object utilizing method shown in Figure 2 to be organized into the leaf node that searches in the image block set of tree construction.As shown in Figure 6, for the query point shown in the figure, be image block 2 with the nearest image block of this query point in the image block set, rather than image block 1.But actual divisional plane possibly be P601, and according to this divisional plane, the leaf node with this query point coupling that inquiry obtains in tree construction is an image block 1.Through utilizing method shown in Figure 3, can one group of pseudo-point be linked with each leaf node (corresponding to an image block), form neighbour's collection of each leaf node.For example, the neighbour of image block 1 collection can comprise image block 2.Like this, further concentrate search, promptly can obtain the image block that matees the most with query point the neighbour.
Particularly; For certain the image block Pt (corresponding to a leaf node in the tree construction) in the image block set; Can at first find the unit Cell_Pt (being also referred to as bounding box, i.e. bounding box) that comprises this image block Pt in the image block ensemble space, as shown in Figure 7.This unit Cell-Pt obtains the minimum space unit that the pairing a plurality of divisional plane of leaf node Pt is surrounded corresponding to cutting apart in the tree construction.Then, in this unit Cell-Pt, generate a proper vector (being called query point) Qt at random.Adopt exhaustive search algorithm for example in the image block set, to search the arest neighbors image block Pt_true of this query point.If the Pt_true that finds is different with image block Pt, then with this Pt_true as a pseudo-point, place the puppet point tabulation of image block Pt.If had this Pt_true in the tabulation of pseudo-point, then the frequency of occurrences count value with correspondence adds 1, otherwise the frequency of occurrences count value of correspondence is set to 1.Repeatedly repeat above-mentioned generated query point at random to the step (concrete number of times can be confirmed according to practical application, does not limit) of upgrading the tabulation of pseudo-point here.At last, with each the pseudo-some ordering in the tabulation of puppet point, select to occur one or more bigger neighbour's collection of count value according to corresponding frequency of occurrences count value as image block Pt.For example, can remove in the pseudo-some tabulation frequency of occurrences count value less than the pseudo-point of certain threshold value F_th.Threshold value F_th can set according to concrete application scenarios, does not also limit here.
Fig. 7 shows the synoptic diagram that utilizes the puppet point that the Cell_Pt corresponding with leaf node find.Wherein, P705 and P706 represent actual divisional plane respectively, and the dark-shaded that actual divisional plane surrounded partly is Cell_Pt, and P703 and P704 represent desirable divisional plane respectively.The pseudo-zone errors of putting of needs that 701 and 702 expressions cause owing to the error between actual divisional plane and the desirable divisional plane; Wherein, The zone of the puppet point that the frequency of occurrences that 701 expressions find according to the method for above-mentioned example is high, 702 represent the zone of the pseudo-point (can remove) that the frequency of occurrences that the method according to above-mentioned example finds is low.
In this example, at first find and leaf node Pt corresponding cells Cell_Pt, therefore, can guarantee that the proper vector (query point) that generates at random all is positioned at this unit, thereby avoid the query point of generation error, guarantee the accuracy of neighbour's collection of generation.In addition, through calculating the frequency of occurrences of each the pseudo-point in the tabulation of pseudo-point, can guarantee only to keep the high pseudo-point of the frequency of occurrences, thereby can improve the speed of storage, and reduce required storage space.
The image processing equipment that utilizes said method also is provided according to embodiment of the present disclosure.
Figure 10 shows the schematic block diagram according to the image processing equipment of an embodiment of the present disclosure.Shown in figure 10, this image processing equipment 1000 can comprise image block collection processing device 1001 and information-storing device 1003.
Image block collection processing device 1001 can be handled the image block set that comprises a plurality of image blocks that from sample image, extract.Particularly, image block collection processing device 1001 can be searched for one or more neighbour's image block of each image block in this image block set in said image block set.Similar with the method embodiment/example of preceding text; Neighbour's image block of each image block that obtains through the searching image set of blocks can be used as neighbour's collection of each image block, and can comprise one or more nearest image block of this image block set each image block of middle distance.
Information-storing device 1003 can preserve image block set and treating apparatus 1001 that obtain with this set in the neighbour of each image block collect relevant information.
Utilize equipment shown in Figure 10, obtain comprising the training set of a plurality of image blocks through training after, can generate neighbour's collection of each image block in this training set.When preserving this training set, can also preserve the wherein information of neighbour's collection of each image block.Like this; In subsequent image processing, need this training set of inquiry fashionable, after inquiry obtained certain image block, the neighbour that can also obtain this image block collected information; And can be as required and concentrate this neighbour and to carry out search refinement further, thereby the image block that is more suited the requirements.With the method or the apparatus in comparison that only generate and preserve the training set, the information that equipment shown in Figure 10 generates more helps follow-up inquiry, promptly can inquire about the information more accurately that obtains.
In certain embodiments, image block collection processing device 1001 can also further be organized into the image block set tree structure of classification.For example, can adopt in the preceding text and the image block set is organized into tree construction, no longer repeat here with reference to the described method of figure 2, Fig. 4 and Fig. 5.
In certain embodiments, image block collection processing device 1001 can also adopt in the preceding text and come the neighbour of each image block in the searching image set of blocks to collect with reference to the described method of figure 3, Fig. 6 and Fig. 7, also no longer repeats here.
In certain embodiments, the image block in the image block set in method/equipment above-mentioned and that hereinafter will describe can be in groups, and every group of image block comprises two or more image blocks that correspond to each other.Like this; A plurality of neighbour's image blocks (like step 103 or 203) of searching for each image block can comprise: one or more the nearest image block of an image block in the set of said image block in search and the every group of image block, that said one or more image block is corresponding one group or more the multiple series of images piece as neighbour's collection of this group image block.In this case, if the image block set is organized into tree construction, then each leaf node in this tree construction is corresponding to set of diagrams picture piece.For example; In some Image Super Resolution Processing; Can generate and comprise many image block set the low resolution that corresponds to each other and high-definition picture piece; Promptly can utilize a large amount of low resolution and high-definition picture sample to extracting many low resolution and high-definition picture pieces to corresponding to each other, formation comprises many training set to image block.When collecting, can only search for neighbour's collection of one of every pair of image block the neighbour of each image block of search, the image block of the corresponding resolution of concentrating with the neighbour of image block promptly can be used as this another neighbour in image block is collected.In other words, every pair of neighbour's collection that image block can corresponding a pair ofly correspond to each other.When utilizing for example single low-resolution image to generate high-definition picture; Can be with certain zone in the low-resolution image as query object; The low-resolution image piece that inquiry matches in the training set; Thereby obtain the high-definition picture piece corresponding, utilize this high-definition picture piece both can rebuild and regional corresponding high-resolution areas as query object with this low-resolution image piece.
Fig. 8 shows the indicative flowchart according to the image processing method of another embodiment of the present disclosure.In this method shown in Figure 8, utilize the set of the method for the example/embodiment in the preceding text or image block that equipment obtains and neighbour's collection information to carry out Flame Image Process.
As shown in Figure 8, this image processing method can comprise step 813 and step 815.
In step 813, for certain query object, can at first in the image block set that comprises a plurality of image blocks that from sample image, extract, carry out coarse search, obtain and the approximate nearest image block of this query object distance.Should be understood that and to adopt any suitable searching method to carry out this coarse search, for example, can adopt random search, Chaos Search or uniformly-spaced search or the like, do not enumerate one by one here.
Each image block in the image block set all has neighbour's collection, and this neighbour's collection comprises one or more nearest image block of said image block set this image block of middle distance.Therefore, behind certain image block that coarse search arrives, can in step 815, carry out search refinement, promptly concentrate and search one or more image block nearest with said query object its neighbour.Search to such an extent that the image block that obtains can be used as one or more arest neighbors image block of said query object through this refinement.
In the method shown in Figure 8, utilized the neighbour of each image block in the image block set that method or equipment in the preceding text generates to collect information.Through utilizing these information, can inquiry be divided into coarse search and two steps of search refinement, promptly earlier search the result with the query object approximate match fast through coarse search, the neighbour according to this result collects to come search refinement again, thereby finds the result of coupling the most.Like this, the speed that can accelerate to inquire about has obtained Query Result accurately simultaneously.
In certain embodiments, the image block set can be organized into hierarchical tree structure, wherein, and one or set of diagrams picture piece in each leaf node correspondence image set of blocks.Root node can the correspondence image set of blocks, and comprises the information that is used for this set is divided into the thresholding of the subclass corresponding with the next stage node.One sub-set of each branch node correspondence image set of blocks and comprising is used for the subclass corresponding with this branch node is divided into the information with the thresholding of the corresponding subclass of next stage node.As a concrete example, the image block set can be organized into structure as shown in Figure 5.In this case; The step (like step 813) of in the set of said image block, carrying out coarse search can comprise: begin to next stage step by step and node motion said query object coupling from the root node of said tree construction; Up to the leaf node of arrival with said query object approximate match; Wherein, in each level, select in the next stage and node said query object coupling according to the information in the node.Utilizing after this coarse search obtains the leaf node of certain and query object approximate match, carrying out search refinement, can obtain the structure of mating the most with query object through further concentrating the neighbour of this leaf node.In the method, need not travel through, and the neighbour of leaf node concentrates the element comprise less relatively entire image set of blocks (tree construction), therefore, when obtaining accurately Query Result, the speed that can also accelerate to inquire about.
Similar with some embodiment in the preceding text, the image block in the image block set can be in groups.Every group of image block can comprise two or more image blocks that correspond to each other.
For example, in some Image Super Resolution Processing, can generate and comprise many image blocks set to image block, every pair of image block can comprise a pair of low-resolution image piece that corresponds to each other and high-definition picture piece.Particularly, can utilize a large amount of low resolution and high-definition picture sample, comprise many training set (or being called dictionary) image block thereby form to extracting many low resolution and high-definition picture pieces to corresponding to each other.When utilizing low-resolution image to generate high-definition picture, can utilize such training set.The low-resolution image that utilizes that Fig. 9 shows according to a specific embodiment of the present disclosure generates the schematic flow of the image super-resolution method of high-definition picture.
As shown in Figure 9, this method can comprise step 913,915 and 919.
Step 913 is similar with step 813, is in the step that comprises many image blocks set carrying out coarse searches to image block.Every pair of image block in the image block set comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece.These image blocks all are to utilize a large amount of low resolution and high-definition picture sample to extracting.In addition, each image block in the image block set all has neighbour's collection.This neighbour's collection can be that method or the equipment described in the employing preceding text generates, and no longer repeats here.In order to utilize low-resolution image to rebuild high-definition picture, can be with certain zone in the low-resolution image as query object.At first, in the image block set, carry out coarse search, obtain image block with this query object approximate match.Similar with the embodiment/example of preceding text, can adopt any appropriate method to carry out coarse search.For example; Be organized under the situation of tree construction in the image block set; Can begin step by step to next stage and node motion said query object coupling from the root node of tree construction, up to the leaf node that arrives with said query object approximate match, wherein; In each level, select in the next stage and node said query object coupling according to the information in the node.Can also adopt additive method (like random search, Chaos Search or uniformly-spaced search etc.) to carry out coarse search, no longer repeat here.
In step 915; Coarse search to the neighbour with the approximate nearest image block of query object concentrate and carry out search refinement; To find one or more wherein nearest image block, as one or more arest neighbors image block of this query object with query object.These arest neighbors image blocks all are the image blocks of low resolution.
In step 919; At first find one or more high-definition picture piece corresponding with one or more arest neighbors image block of query object; Then; Utilize these high-definition picture pieces to generate the high-definition picture zone, as in the high-definition picture with the corresponding part in said query object position.Can adopt any appropriate method to utilize one or more high-definition picture piece to generate said high-definition picture zone; For example can adopt the correlation technique in the following document (being called pertinent literature 1): " Image Hallucination Using Neighbor Embedding Over Visual Primitive Manifolds " (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR) of people such as Wei Fan; The Minn., 18-23 day in June, 2007).The content of the document is herein incorporated by reference.
Like this, above processing is all carried out in each zone in the low-resolution image, can be rebuild the high-definition picture corresponding with this low-resolution image.In the method shown in Figure 9, utilized the neighbour of each image block in the image block set that method or equipment in the preceding text generates to collect information equally.Like this, the speed that can accelerate to inquire about has obtained Query Result accurately simultaneously.In addition, adopt the method for Fig. 9, can only utilize the single width low-resolution image to generate high-definition picture.
Figure 11 shows the schematic block diagram according to the image processing equipment of an embodiment of the present disclosure.This image processing equipment 1100 is similar with method shown in Figure 8, has utilized the set of the method for the example/embodiment in the preceding text or image block that equipment obtains and neighbour's collection information to carry out Flame Image Process.
Shown in figure 11, image processing equipment 1100 can comprise coarse search device 1102 and search device for thinning 1104.
For certain query object, coarse search device 1102 can carry out coarse search in the image block set that comprises a plurality of image blocks that from sample image, extract, and obtains and the approximate nearest image block of this query object distance.Should be understood that coarse search device 1102 can adopt any suitable searching method to carry out this coarse search, no longer repeat here.
Each image block in the image block set all has neighbour's collection, and this neighbour's collection comprises one or more nearest image block of said image block set this image block of middle distance.Therefore; Behind certain image block that coarse search device 1102 searches; Search device for thinning 1104 can be concentrated the neighbour of this image block and search one or more image block nearest with said query object, as one or more arest neighbors image block of said query object.
Equipment utilization shown in Figure 11 the neighbour of each image block in the image block set that generates of method or the equipment in the preceding text collect information, the speed that can accelerate to inquire about obtains Query Result accurately simultaneously.
Similar with some embodiment in the preceding text, the image block set can be organized into hierarchical tree structure, wherein, and one or set of diagrams picture piece in each leaf node correspondence image set of blocks.Root node can the correspondence image set of blocks, and comprises the information that is used for this set is divided into the thresholding of the subclass corresponding with the next stage node.One sub-set of each branch node correspondence image set of blocks and comprising is used for the subclass corresponding with this branch node is divided into the information with the thresholding of the corresponding subclass of next stage node.As a concrete example, the image block set can be organized into structure as shown in Figure 5.In this case; Coarse search device 1102 can begin step by step to next stage and node motion said query object coupling from the root node of said tree construction; Up to the leaf node of arrival with said query object approximate match; Wherein, in each level, select in the next stage and node said query object coupling according to the information in the node.Promptly can be used as the output result of coarse search device 1102 with the leaf node of query object approximate match.Need not travel through like this, and the element that the neighbour of leaf node collection comprises is less relatively, therefore, when obtaining accurately Query Result, can also further accelerates the speed of inquiring about entire image set of blocks (tree construction).
Similar with some embodiment in the preceding text, the image block in the image block set can be in groups.Every group of image block can comprise two or more image blocks that correspond to each other.
Figure 12 shows the schematic block diagram according to the image processing equipment of another specific embodiment of the present disclosure.This equipment 1200 is similar with equipment 1100; Difference is; In this embodiment, the image block in the image block set that is utilized is paired, and every pair of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece; In addition, each image block in the image block set all has neighbour's collection.This neighbour's collection can be that method or the equipment described in the employing preceding text generates, and no longer repeats here.Shown in figure 12, except coarse search device 1202 and search device for thinning 1204, equipment 1200 also comprises resolution intensifier 1206.
Search device for thinning 1204 device 1202 coarse searches to the neighbour with the approximate nearest image block of query object concentrate and carry out search refinement; To find one or more wherein nearest image block, as one or more arest neighbors image block of this query object with query object.These arest neighbors image blocks all are the image blocks of low resolution.
Then; Resolution intensifier 1206 can at first find one or more high-definition picture piece corresponding with one or more arest neighbors image block of query object; Then; Utilize these high-definition picture pieces to generate the high-definition picture zone, as in the high-definition picture with the corresponding part in said query object position.Resolution intensifier 1206 can adopt any appropriate method (method that the pertinent literature of for example mentioning in the preceding text 1 relates to) to utilize one or more high-definition picture piece to generate said high-definition picture zone, no longer repeats here.Like this, above processing is all carried out in each zone in 1202,1204 and 1206 pairs of low-resolution images of use device, can rebuild the high-definition picture corresponding with this low-resolution image.
Similar with method embodiment shown in Figure 9, that equipment shown in Figure 12 has utilized equally is that method or the equipment in the preceding text generates, the neighbour of each image block in the image block set collects information.Like this, the speed that can accelerate to inquire about has obtained Query Result accurately simultaneously.In addition, the equipment of Figure 12 can only utilize the single width low-resolution image to generate high-definition picture.
Figure 13 (a) shows a width of cloth low-resolution image.Figure 13 (b) show utilize bilinear interpolation that generate with the corresponding high-definition picture of low-resolution image Figure 13 (a).Figure 13 (c) show utilization with reference to the method for figure 9 or embodiment of the present disclosure shown in Figure 12 or equipment that generate with the corresponding high-definition picture of low-resolution image Figure 13 (a).Figure 13 (d) shows true shooting and the corresponding high-definition picture of low-resolution image Figure 13 (a).Can find out that it is gratifying utilizing the quality of the high-definition picture that method or equipment according to embodiment of the present disclosure generates.
Should be understood that the foregoing description and example are exemplary, rather than exhaustive, the disclosure should not be regarded as and be confined to any concrete embodiment or example.In embodiment of the present disclosure or example, related image pattern or image block can be any forms, do not do any qualification here.
As an example, each composition module of each step of said method and the said equipment and/or software, firmware, hardware or its combination that device may be embodied as.For example, can be used as the part of various device (, not enumerating one by one) here like computing machine, personal digital assistant, camera, video camera and other image processing equipments etc. with image processing function.Spendable concrete means or mode were well known to those skilled in the art when each was formed the mode of module through software, firmware, hardware or its combination and is configured in the said equipment, repeated no more at this.
As stated, in said method and the equipment image is carried out processed steps and module can realize through software.To the computing machine with specialized hardware structure (multi-purpose computer 1400 for example shown in Figure 14) program that constitutes this software is installed from storage medium or network, this computing machine can be carried out various functions etc. when various program is installed.
In Figure 14, CPU (CPU) 1401 carries out various processing according to program stored among ROM (read-only memory) (ROM) 1402 or from the program that storage area 1408 is loaded into random-access memory (ram) 1403.In RAM 1403, also store data required when CPU 1401 carries out various processing or the like as required.CPU 1401, ROM 1402 and RAM 1403 are connected to each other via bus 1404.Input/output interface 1405 also is connected to bus 1404.
Following parts are connected to input/output interface 1405: importation 1406 (comprising keyboard, mouse or the like), output 1407 (comprise display; Such as cathode ray tube (CRT), LCD (LCD) etc. and loudspeaker etc.), storage area 1408 (comprising hard disk etc.), communications portion 1409 (comprising that NIC is such as LAN card, modulator-demodular unit etc.).Communications portion 1409 is handled such as the Internet executive communication via network.As required, driver 1410 also can be connected to input/output interface 1405.Detachable media 1411 is installed on the driver 1410 such as disk, CD, magneto-optic disk, semiconductor memory or the like as required, makes the computer program of therefrom reading be installed to as required in the storage area 1408.
Realizing through software under the situation of above-mentioned series of processes, such as detachable media 1411 program that constitutes software is being installed such as the Internet or storage medium from network.
It will be understood by those of skill in the art that this storage medium is not limited to shown in Figure 14 wherein having program stored therein, distribute so that the detachable media 1411 of program to be provided to the user with equipment with being separated.The example of detachable media 1411 comprises disk (comprising floppy disk (registered trademark)), CD (comprising compact disc read-only memory (CD-ROM) and digital universal disc (DVD)), magneto-optic disk (comprising mini-disk (MD) (registered trademark)) and semiconductor memory.Perhaps, storage medium can be hard disk that comprises in ROM 1402, the storage area 1408 or the like, computer program stored wherein, and be distributed to the user with the equipment that comprises them.
The disclosure also proposes a kind of program product that stores the instruction code of machine-readable.When said instruction code is read and carried out by machine, can carry out above-mentioned method according to disclosure embodiment.
Correspondingly, the storage medium that is used for carrying the program product of the above-mentioned instruction code that stores machine-readable is also included within of the present disclosure open.Said storage medium includes but not limited to floppy disk, CD, magneto-optic disk, storage card, memory stick or the like.
In the above in the description to disclosure specific embodiment; Characteristic to a kind of embodiment is described and/or illustrated can use identical or similar mode in one or more other embodiment, to use; Combined with the characteristic in other embodiment, or substitute the characteristic in other embodiment.
Should stress that term " comprises/comprise " existence that when this paper uses, refers to characteristic, key element, step or assembly, but not get rid of the existence of one or more further feature, key element, step or assembly or additional.
In addition, the time sequencing of describing during method of the present disclosure is not limited to is to specifications carried out, also can according to other time sequencing ground, carry out concurrently or independently.The execution sequence of the method for therefore, describing in this instructions does not constitute restriction to technical scope of the present disclosure.
Can find out through above description,, following scheme is provided according to embodiment of the present disclosure:
1. 1 kinds of image processing methods of remarks comprise:
In the image block set that comprises a plurality of image blocks that from sample image, extract; Search for one or more neighbour's image block of each image block; Wherein, Said one or more neighbour's image block comprises one or more nearest image block of said image block set this each image block of middle distance, thereby forms neighbour's collection of this each image block; And
Preserve the information of neighbour's collection of said image block set and relevant each image block.
Remarks 2. is like remarks 1 described method, and wherein, a plurality of neighbour's image blocks of searching for each image block comprise:
Generate a proper vector at random, wherein the distance of this proper vector and this image block is in predetermined scope;
In said image block set, search the nearest arest neighbors image block of the said proper vector of distance; If the arest neighbors image block that finds is different from this image block; Then increase the related count value of this arest neighbors image block that finds; Wherein, generate at least two proper vectors and search the arest neighbors image block of each proper vector at random; And
With one or more that find, related arest neighbors image block, as neighbour's collection of this image block with bigger count value.
One or more nearest image block of an image block in the set of said image block in search and the every group of image block, that said one or more image block is corresponding one group or more the multiple series of images piece as neighbour's collection of this group image block.
Remarks 4. is like remarks 3 described methods, and wherein, every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece.
Remarks 5. is like each described method among the remarks 1-4; Also comprise: said image block set is organized into multistage tree construction; Wherein, The corresponding said image block set of root node in the said tree construction, and comprise the information that is used for said set is divided into the thresholding of the subclass corresponding with the next stage node; One sub-set of the corresponding said image block set of each branch node and comprising is used for the subclass corresponding with this branch node is divided into the information with the thresholding of the corresponding subclass of next stage node; And the set of diagrams picture piece in the corresponding said image block set of each leaf node.
Remarks 6. is like remarks 5 described methods, and wherein, the image block in the said image block set is in groups, and every group of image block comprise two or more image blocks that correspond to each other, and said image block set is organized into multistage tree construction comprises:
Recurrence is cut apart said image block set, wherein, in cutting apart multiple series of images piece to be split is divided into two sub-set at every turn, in each subclass, comprises set of diagrams picture piece.
7. 1 kinds of image processing methods of remarks comprise:
In the image block set that comprises a plurality of image blocks that from sample image, extract, carry out coarse search, obtain and the approximate nearest image block of query object distance; And
The neighbour of the image block that arrives at coarse search concentrates; Search one or more image block nearest with said query object; One or more arest neighbors image block as said query object; Wherein, coarse search to the neighbour collection of image block comprise one or more nearest image block of said image block set this image block of middle distance.
Remarks 8. is like remarks 7 described methods; Wherein, Image block in the said image block set is in groups; Every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece, and said query object is certain zone in the low-resolution image, and said method also comprises:
Utilize one or more high-definition picture piece corresponding to generate the high-definition picture zone with one or more arest neighbors image block of said query object, as in the high-definition picture with the corresponding part in said query object position.
Remarks 9. is like remarks 7 or 8 described methods; Wherein, The set of said image block is organized into hierarchical tree structure, and one or set of diagrams picture piece in the corresponding said image block set of each leaf node of said tree construction, and in said image block set, carries out coarse search and comprise:
Begin to next stage step by step and node motion said query object coupling from the root node of said tree construction; Up to arriving the leaf node that matees with said query object; Wherein, in each level, select in the next stage and node said query object coupling according to the information in the node.
10. 1 kinds of image processing equipments of remarks comprise:
The image block collection processing device; Be used for the image block set that comprises a plurality of image blocks that extract from sample image; Search for one or more neighbour's image block of each image block; Wherein, said one or more neighbour's image block comprises one or more nearest image block of said image block set this each image block of middle distance, thereby forms neighbour's collection of this each image block; And
Information-storing device is used to preserve the information that said image block is gathered and the neighbour of relevant each image block collects.
Remarks 11. is like remarks 10 described equipment, and wherein, said image block collection processing device is arranged to through a plurality of neighbour's image blocks with each image block of search that gets off:
Generate a proper vector at random, wherein the distance of this proper vector and this image block is in predetermined scope;
In said image block set, search the nearest arest neighbors image block of the said proper vector of distance; If the arest neighbors image block that finds is different from this image block; Then increase and this related count value of arest neighbors image block that finds; Wherein, generate at least two proper vectors and search the arest neighbors image block of each proper vector at random; And
With one or more that find, related arest neighbors image block, as neighbour's collection of this image block with bigger count value.
Remarks 12. is like remarks 10 described equipment, and wherein, the image block in the said image block set is in groups, and every group of image block comprise two or more image blocks that correspond to each other, and said image block collection processing device is arranged to:
One or more nearest image block of an image block in the set of said image block in search and the every group of image block, that said one or more image block is corresponding one group or more the multiple series of images piece as neighbour's collection of this group image block.
Remarks 13. is like remarks 12 described equipment, and wherein, every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece.
Remarks 14. is like each described equipment among the remarks 10-13; Wherein, Said image block collection processing device also is arranged to said image block set is organized into multistage tree construction; Wherein, the corresponding said image block set of the root node in the said tree construction, and comprise the information that is used for said set is divided into the thresholding of the subclass corresponding with the next stage node; One sub-set of the corresponding said image block set of each branch node, and comprise and be used for the subclass corresponding with this branch node is divided into the information with the thresholding of the corresponding subclass of next stage node; And the set of diagrams picture piece in the corresponding said image block set of each leaf node.
Remarks 15. is like remarks 14 described methods; Wherein, Image block in the said image block set is in groups; And every group of image block comprises two or more image blocks that correspond to each other, and said image block collection processing device is arranged to through to get off said image block set being organized into multistage tree construction:
Recurrence is cut apart said image block set, wherein, in cutting apart multiple series of images piece to be split is divided into two sub-set at every turn, in each subclass, comprises set of diagrams picture piece.
16. 1 kinds of image processing equipments of remarks comprise:
The coarse search device is used for the image block set that comprises a plurality of image blocks that extract from sample image, carrying out coarse search, obtains and the approximate nearest image block of query object distance; And
The search device for thinning; The neighbour of the image block that is used for arriving at said coarse search device coarse search concentrates; Search one or more image block nearest with said query object; As one or more arest neighbors image block of said query object, wherein, coarse search to the neighbour collection of image block comprise one or more nearest image block of said image block set this image block of middle distance.
Remarks 17. is like remarks 16 described equipment; Wherein, Image block in the said image block set is in groups; Every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece, and said query object is certain zone in the low-resolution image, and said equipment also comprises:
The resolution intensifier; Be used to utilize one or more high-definition picture piece corresponding to generate the high-definition picture zone with one or more arest neighbors image block of said query object, as in the high-definition picture with the corresponding part in said query object position.
Although the disclosure is disclosed above through description to specific embodiment of the present disclosure,, should be appreciated that all above-mentioned embodiment and example all are exemplary, and nonrestrictive.Those skilled in the art can be in the spirit of accompanying claims and scope design to various modifications of the present disclosure, improve or equivalent.These modifications, improvement or equivalent also should be believed to comprise in protection domain of the present disclosure.
Claims (10)
1. image processing method comprises:
In the image block set that comprises a plurality of image blocks that from sample image, extract; Search for one or more neighbour's image block of each image block; Wherein, Said one or more neighbour's image block comprises one or more nearest image block of said image block set this each image block of middle distance, thereby forms neighbour's collection of this each image block; And
Preserve the information of neighbour's collection of said image block set and relevant each image block.
2. a plurality of neighbour's image blocks of the method for claim 1, wherein searching for each image block comprise:
Generate a proper vector at random, wherein the distance of this proper vector and this image block is in predetermined scope;
In said image block set, search the nearest arest neighbors image block of the said proper vector of distance; If the arest neighbors image block that finds is different from this image block; Then increase and this related count value of arest neighbors image block that finds; Wherein, generate at least two proper vectors and search the arest neighbors image block of each proper vector at random; And
With one or more that find, related arest neighbors image block, as neighbour's collection of this image block with bigger count value.
3. the image block in the set of the method for claim 1, wherein said image block is in groups, and every group of image block comprise two or more image blocks that correspond to each other, and a plurality of neighbour's image blocks of searching for each image block comprise:
One or more nearest image block of an image block in the set of said image block in search and the every group of image block, that said one or more image block is corresponding one group or more the multiple series of images piece as neighbour's collection of this group image block.
4. image processing method comprises:
In the image block set that comprises a plurality of image blocks that from sample image, extract, carry out coarse search, obtain and the approximate nearest image block of query object distance; And
The neighbour of the image block that arrives at coarse search concentrates; Search one or more image block nearest with said query object; One or more arest neighbors image block as said query object; Wherein, coarse search to the neighbour collection of image block comprise one or more nearest image block of said image block set this image block of middle distance.
5. method as claimed in claim 4; Wherein, Image block in the said image block set is in groups; Every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece, and said query object is certain zone in the low-resolution image, and said method also comprises:
Utilize one or more high-definition picture piece corresponding to generate the high-definition picture zone with one or more arest neighbors image block of said query object, as in the high-definition picture with the corresponding part in said query object position.
6. image processing equipment comprises:
The image block collection processing device; Be used for the image block set that comprises a plurality of image blocks that extract from sample image; Search for one or more neighbour's image block of each image block; Wherein, said one or more neighbour's image block comprises one or more nearest image block of said image block set this each image block of middle distance, thereby forms neighbour's collection of this each image block; And
Information-storing device is used to preserve the information that said image block is gathered and the neighbour of relevant each image block collects.
7. equipment as claimed in claim 6, wherein, said image block collection processing device is arranged to through a plurality of neighbour's image blocks with each image block of search that gets off:
Generate a proper vector at random, wherein the distance of this proper vector and this image block is in predetermined scope;
In said image block set, search the nearest arest neighbors image block of the said proper vector of distance; If the arest neighbors image block that finds is different from this image block; Then increase and this related count value of arest neighbors image block that finds; Wherein, generate at least two proper vectors and search the arest neighbors image block of each proper vector at random; And
With one or more that find, related arest neighbors image block, as neighbour's collection of this image block with bigger count value.
8. equipment as claimed in claim 6, wherein, the image block in the set of said image block is in groups, and every group of image block comprise two or more image blocks that correspond to each other, and said image block collection processing device is arranged to:
One or more nearest image block of an image block in the set of said image block in search and the every group of image block, that said one or more image block is corresponding one group or more the multiple series of images piece as neighbour's collection of this group image block.
9. image processing equipment comprises:
The coarse search device is used for the image block set that comprises a plurality of image blocks that extract from sample image, carrying out coarse search, obtains and the approximate nearest image block of query object distance; And
The search device for thinning; The neighbour of the image block that is used for arriving at said coarse search device coarse search concentrates; Search one or more image block nearest with said query object; As one or more arest neighbors image block of said query object, wherein, coarse search to the neighbour collection of image block comprise one or more nearest image block of said image block set this image block of middle distance.
10. equipment as claimed in claim 9; Wherein, Image block in the said image block set is in groups; Every group of image block comprises a pair of low-resolution image piece that corresponds to each other and high-definition picture piece, and said query object is certain zone in the low-resolution image, and said equipment also comprises:
The resolution intensifier; Be used to utilize one or more high-definition picture piece corresponding to generate the high-definition picture zone with one or more arest neighbors image block of said query object, as in the high-definition picture with the corresponding part in said query object position.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010257628.1A CN102375990B (en) | 2010-08-17 | 2010-08-17 | Method and equipment for processing images |
| JP2011178595A JP2012043437A (en) | 2010-08-17 | 2011-08-17 | Image processing method and image processing device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010257628.1A CN102375990B (en) | 2010-08-17 | 2010-08-17 | Method and equipment for processing images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102375990A true CN102375990A (en) | 2012-03-14 |
| CN102375990B CN102375990B (en) | 2015-03-04 |
Family
ID=45794568
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010257628.1A Expired - Fee Related CN102375990B (en) | 2010-08-17 | 2010-08-17 | Method and equipment for processing images |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP2012043437A (en) |
| CN (1) | CN102375990B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500176A (en) * | 2013-09-06 | 2014-01-08 | 清华大学 | Piece net model and construction method thereof |
| CN103970769A (en) * | 2013-01-29 | 2014-08-06 | 华为技术有限公司 | Image retrieval method and image retrieval device |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6278042B2 (en) | 2013-04-05 | 2018-02-14 | 日本電気株式会社 | Information processing apparatus and image processing method |
| JP6800453B1 (en) * | 2020-05-07 | 2020-12-16 | 株式会社 情報システムエンジニアリング | Information processing device and information processing method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6226420B1 (en) * | 1997-12-26 | 2001-05-01 | International Business Machines Corporation | Non-integer scaling of raster images |
| CN1378159A (en) * | 2001-03-26 | 2002-11-06 | Lg电子株式会社 | Picture searching method and device |
| US7474337B1 (en) * | 2000-10-24 | 2009-01-06 | Sony Corporation | Method and apparatus to provide edge enhancements as part of a demosaicing process |
| CN101482925A (en) * | 2009-01-16 | 2009-07-15 | 西安电子科技大学 | Photograph generation method based on local embedding type hidden Markov model |
| CN101556601A (en) * | 2009-03-12 | 2009-10-14 | 华为技术有限公司 | Method and device for searching k neighbor |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4883649B2 (en) * | 2006-08-31 | 2012-02-22 | 公立大学法人大阪府立大学 | Image recognition method, image recognition apparatus, and image recognition program |
-
2010
- 2010-08-17 CN CN201010257628.1A patent/CN102375990B/en not_active Expired - Fee Related
-
2011
- 2011-08-17 JP JP2011178595A patent/JP2012043437A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6226420B1 (en) * | 1997-12-26 | 2001-05-01 | International Business Machines Corporation | Non-integer scaling of raster images |
| US7474337B1 (en) * | 2000-10-24 | 2009-01-06 | Sony Corporation | Method and apparatus to provide edge enhancements as part of a demosaicing process |
| CN1378159A (en) * | 2001-03-26 | 2002-11-06 | Lg电子株式会社 | Picture searching method and device |
| CN101482925A (en) * | 2009-01-16 | 2009-07-15 | 西安电子科技大学 | Photograph generation method based on local embedding type hidden Markov model |
| CN101556601A (en) * | 2009-03-12 | 2009-10-14 | 华为技术有限公司 | Method and device for searching k neighbor |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103970769A (en) * | 2013-01-29 | 2014-08-06 | 华为技术有限公司 | Image retrieval method and image retrieval device |
| CN103970769B (en) * | 2013-01-29 | 2018-06-26 | 华为技术有限公司 | Image search method and device |
| CN103500176A (en) * | 2013-09-06 | 2014-01-08 | 清华大学 | Piece net model and construction method thereof |
| CN103500176B (en) * | 2013-09-06 | 2016-08-31 | 清华大学 | Sheet pessimistic concurrency control and construction method thereof |
| US9836880B2 (en) | 2013-09-06 | 2017-12-05 | Tsinghua University | Patch net model and construction method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102375990B (en) | 2015-03-04 |
| JP2012043437A (en) | 2012-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chandio et al. | Precise single-stage detector | |
| US11514063B2 (en) | Method and apparatus of recommending information based on fused relationship network, and device and medium | |
| CN102194149B (en) | Community discovery method | |
| CN107748745B (en) | Enterprise name keyword extraction method | |
| CN102129560A (en) | Method and device for identifying characters | |
| CN113971209A (en) | Non-supervision cross-modal retrieval method based on attention mechanism enhancement | |
| Jung et al. | Devil's on the edges: Selective quad attention for scene graph generation | |
| CN104200272A (en) | Complex network community mining method based on improved genetic algorithm | |
| CN102375990A (en) | Method and equipment for processing images | |
| CN104573683B (en) | Character string identification method and device | |
| CN111144300B (en) | Pdf table structure identification method based on image identification | |
| CN112165401A (en) | Edge community discovery algorithm based on network pruning and local community expansion | |
| CN110263021B (en) | Theme library generation method based on personalized label system | |
| CN105956158A (en) | Automatic extraction method of network neologism on the basis of mass microblog texts and use information | |
| CN105404677A (en) | Tree structure based retrieval method | |
| Luqman et al. | Subgraph spotting through explicit graph embedding: An application to content spotting in graphic document images | |
| CN114677357A (en) | Model, method and equipment for detecting self-explosion defect of aerial photographing insulator and storage medium | |
| CN116186310B (en) | AR space labeling and displaying method fused with AI general assistant | |
| Zhao et al. | ProsegDL: Binary protocol format extraction by deep learning-based field boundary identification | |
| CN117312681A (en) | Meta universe oriented user preference product recommendation method and system | |
| CN107463679A (en) | A kind of information recommendation method and device | |
| CN107729898B (en) | Method and device for detecting text lines in text image | |
| CN104463864A (en) | Multistage parallel key frame cloud extraction method and system | |
| US20170103285A1 (en) | Method and device for detecting copies in a stream of visual data | |
| CN111221993B (en) | Visual media retrieval method based on depth binary detail perception hash |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150304 Termination date: 20180817 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |