CN102479379B - Image rectification method and relevant image rectification system - Google Patents
Image rectification method and relevant image rectification system Download PDFInfo
- Publication number
- CN102479379B CN102479379B CN201010558117.3A CN201010558117A CN102479379B CN 102479379 B CN102479379 B CN 102479379B CN 201010558117 A CN201010558117 A CN 201010558117A CN 102479379 B CN102479379 B CN 102479379B
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- correction
- pixels
- corrected
- 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.)
- Active
Links
Abstract
The invention provides an image rectification method and a relevant image rectification system, which can be used to perform rectification on an image captured by a fisheye lens or ultra wide angle lens, to reduce a geometric distortion phenomenon in the image, and can perform geometric adjustment on an image according to user's requirements. The invention enhances image rectification processing efficiency via a memory configuration technique.
Description
Technical field
The present invention relates to image and process, relate in particular to a kind of in order to proofread and correct because of fish eye lens and the image processing method of the caused geometric distortion of bugeye lens and relevant image processing system.
Background technology
Fish eye lens and bugeye lens are widely used in monitoring camera or general digital camera is medium.Yet, although the scenery in the camera lens fechtable of this type in quite with great visual angle, but its formed image is often accompanied by serious geometric distortion (geometrical distortion), these geometric distortions can users cannot be easily and some details in image that picks out clearly, or even cannot identification.Therefore, the image generating by fish eye lens and bugeye lens must pass through suitable image disposal skill, just can restore the image that comparatively meets original scenery, in order to user, observes.
Summary of the invention
Therefore, the invention provides a kind of method for correcting image and relevant corrective system, it is in order to the picture capturing by fish eye lens or bugeye lens is proofreaied and correct, to alleviate the phenomenon of the geometric distortion in picture.Moreover image processing method of the present invention uses a kind of storer configuration skill of reference data alignment characteristics to relevant image processing system, and then promotes the processing speed of promoting the interpolative operation in image rectification.In addition, need pass through coordinate transformation method owing to processing geometric distortion, the parameter that the present invention also uses coordinate transformation method proposes a kind of method for parameter estimation, by method for parameter estimation proposed by the invention, and the complex process can reduced parameter obtaining.
According to one embodiment of the invention, it provides a kind of method for correcting image, this method for correcting image in order to according to an image to be corrected to export image after a correction, wherein this corrections afterwards image include block of pixels after a plurality of corrections.Moreover this method for correcting image carries out repeatedly calculated for pixel values operation, wherein, each calculated for pixel values operation is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct.This calculated for pixel values operation includes: set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected; According to this coordinate transformation relation, decision is proofreaied and correct a block of pixels to be corrected of rear block of pixels corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected, wherein, this pending block of pixels has the pending pixel of multiple row (row); The pixel value of each pending pixel is kept in respectively to having in a storer of a plurality of memory banks (memory bank), and the pixel value of the pending pixel of its different lines can be temporary in respectively in the different memory storehouse in this storer; And, utilize after this correction corresponding this second coordinate figure of pixel to determine the corresponding a plurality of specific pending pixels of pixel after this corrections, and a plurality of pixel values that this storer reads the plurality of specific pending pixel certainly carry out an interpolative operation to produce this correction pixel value of pixel afterwards.Wherein, the plurality of pixel value of the plurality of specific pending pixel is kept in to different memory banks.
According to another embodiment of the present invention, it provides a kind of method for correcting image, this method for correcting image in order to according to an image to be corrected to export image after a correction, wherein this corrections afterwards image include block of pixels after a plurality of corrections, and this image to be corrected captures by a fish eye lens.Moreover this method for correcting image carries out repeatedly calculated for pixel values operation, each calculated for pixel values operation is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct.This calculated for pixel values operation includes: set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected, wherein, this coordinate transformation relation is at least determined by the conversion of a perspective distortion (perspective distortion) calibration coordinate, and in this perspective distortion calibration coordinate conversion, corresponding to the angle parameter between an object plane (Object plane) and a picture plane (Image plane), an incident angle with the corresponding incident ray of this image to be corrected decides, wherein, this incident light linear system penetrates this fish-eye photocentre (optical center, according to this coordinate transformation relation, determine a block of pixels to be corrected of block of pixels after proofreading and correct corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected, and, utilize rear corresponding this second coordinate figure of pixel of this correction to determine the rear corresponding a plurality of specific pending pixels of pixel of this correction, and utilize a plurality of pixel values of the plurality of specific pending pixel to carry out an interpolative operation to produce a pixel value of the rear pixel of this correction.
According to another embodiment of the present invention, it provides a kind of image correction system, this image correction system in order to according to an image to be corrected to export image after a correction.Wherein, image includes block of pixels after a plurality of corrections after this correction, and this image correction system carries out repeatedly calculated for pixel values operation, and each calculated for pixel values operation is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct.This image correction system includes: a coordinate transformation relation produces circuit, a block selecting circuit, a storer and an interpolative operation circuit.This coordinate transformation relation produces circuit in order to set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, makes one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected.This block selecting circuit is coupled to this coordinate transformation relation and produces circuit, and in order to according to this coordinate transformation relation, decision is proofreaied and correct a block of pixels to be corrected of rear block of pixels corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected.This storer is coupled to this block selecting circuit, there are a plurality of memory banks (memory bank), and the pixel value of each pending pixel is temporary in respectively in the plurality of memory bank, and wherein this pending block of pixels has the pending pixel of multiple row (row).This interpolative operation unit is coupled to this storer and this coordinate transformation relation produces circuit, and this interpolative operation unary system and utilize after this correction corresponding this second coordinate figure of pixel to determine the corresponding a plurality of specific pending pixels of pixel after this correction, and a plurality of pixel values that this storer reads the plurality of specific pending pixel certainly carry out an interpolative operation to produce this corrections pixel value of pixel afterwards.Wherein, the pixel value of the pending pixel of different lines can be temporary in respectively in the different memory storehouse in this storer.
According to an embodiment more of the present invention, it provides a kind of image correction system, this image correction system in order to according to an image to be corrected to export image after a correction.Wherein, image includes block of pixels after a plurality of corrections after this correction, and this image to be corrected captures by a fish eye lens.Moreover, this image correction system is in order to carry out repeatedly calculated for pixel values operation, the operation of each calculated for pixel values is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct, and this image correction system includes: a coordinate transformation relation produces circuit, a block selecting circuit and an interpolative operation circuit.This coordinate transformation relation produces circuit in order to set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected, wherein, this coordinate transformation relation produces circuit and comprises at least one perspective distortion (perspective distortion) calibration coordinate change-over circuit, and this coordinate transformation relation is at least determined by this perspective distortion calibration coordinate change-over circuit, and in this perspective distortion calibration coordinate change-over circuit, use corresponding to the angle parameter between an object plane (Object plane) and picture plane (Image plane), an incident angle with the corresponding incident ray of this image to be corrected decides, wherein, this incident light linear system penetrates this fish-eye photocentre (optical center.This block selecting circuit is coupled to this coordinate transformation relation and produces circuit, and in order to according to this coordinate transformation relation, decision is proofreaied and correct a block of pixels to be corrected of rear block of pixels corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected.This interpolative operation unit is coupled to storer and this coordinate transformation relation produces circuit, and, corresponding this second coordinate figure of pixel to be to determine the corresponding a plurality of specific pending pixels of pixel after this corrections after this this correction of interpolative operation unit by using, and utilizes a plurality of pixel values of the plurality of specific pending pixel to carry out an interpolative operation to produce this correction pixel value of pixel afterwards.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of an embodiment of method for correcting image of the present invention.
Fig. 2 be illustrate image to be corrected in the present invention and proofread and correct after coordinate transformation relation between image.
Fig. 3 illustrates storer configuration mode of the present invention.
Fig. 4 illustrates image to be corrected and the relation of proofreading and correct rear image in the present invention.
Fig. 5 is the deciding means that illustrates coordinate transformation relation in embodiments of the invention.
Fig. 6 is the formation reason that illustrates perspective distortion.
Fig. 7 is the principle that illustrates parameter estimation mode of the present invention.
Fig. 8 is the function block schematic diagram of an embodiment of image correction system of the present invention.
[main element symbol description]
101～107 steps
201,203 dotted line frames
202 solid box
2031～2034 pixel columns
300,803 storeies
600 lens
700 fish eye lenses
800 image correction systems
801 coordinate transformation relations produce circuit
802 block selecting circuits
804 interpolative operation circuit
810 image sources
Embodiment
Image rectification final purpose of the present invention is to reduce the geometric distortion in the image to be corrected capture by fish eye lens or bugeye lens, to produce image after a correction.Wherein, before producing the rear image of this correction, after this correction, the picture dimension of image must predetermine (but still can change according to different demands), and again image after this correction is divided into block of pixels after the correction that a plurality of sizes are identical, and by the method for interpolative operation, produce one by one each and proofread and correct in rear block of pixels, the pixel value of each pixel (that is: GTG value), last just can be according to all pixel values that calculate, carry out image after output calibration.
For instance, image rectification of the present invention may be redefined for 640*480 pixel by the picture dimension of image after a correction, then again image after this correction is divided into the rear block of pixels of 8*8 correction that size is identical, therefore, wherein each block of pixels to be corrected can comprise 80*60 pixel and (please notes, cited numerical value is only the use of explanation herein, is not restriction of the present invention).And proofread and correct in rear image in this, producing each interpolative operation of proofreading and correct the pixel value of rear pixel needs to find out the relevant position of its script in image to be corrected (by coordinate conversion, coordinate figure), and according to this relevant position, find out a plurality of pixel values of a plurality of pixels to be corrected that are adjacent to this relevant position, recycle these pixel values and carry out interpolative operation to obtain the pixel value of the rear pixel of this correction.Therefore, image rectification of the present invention must carry out repeatedly calculated for pixel values operation, wherein, after each calculated for pixel values operation meeting generation one correction, in block of pixels, each proofreaies and correct the pixel value of rear pixel, the numerical value of enumerating in take is above row, and calculated for pixel values operation each time will obtain 80x60 the pixel value of proofreading and correct rear pixel.About further illustrating of these operations, please refer to Fig. 1, it is the process flow diagram of the calculated for pixel values operation in method for correcting image of the present invention.
Note that the process flow diagram shown in Fig. 1 only represents a possibility embodiment of the calculated for pixel values operation of method for correcting image of the present invention; In other possibility embodiments of the present invention, also may comprise other steps in addition of the step shown in Fig. 1.
First, in step 101, calculated for pixel values operation of the present invention can be set up a coordinate transformation relation between this image to be corrected and the rear image of this correction, make one first coordinate figure (Xc of the rear pixel of each correction of the rear block of pixels of this correction, Yc) correspond to one second coordinate figure (Xd, Yd) in this image to be corrected.For instance, if take Fig. 2 as example, after proofreading and correct in image one proofreaied and correct a plurality of boundary pixels (pixel that dotted line frame 201 comprises) of rear block of pixels A, can obtain its relevant coordinate (coordinate points in solid box 202) on image to be corrected by coordinate transformation relation.In other words, the first coordinate figure (Xc, Yc) value that each that the dotted line frame 201 in image comprises after correction is proofreaied and correct rear pixel can map to one second coordinate figure (Xd, Yd) in this image to be corrected.From another perspective, the second coordinate figure (Xd in this image to be corrected, Yd) image on, to reduce by this coordinate transformation relation, and this image can be based upon the first coordinate figure (Xc in the rear image of this correction, Yc), on position, therefore, coordinate transformation relation will determine the improvement mode of geometric distortion.
Then, after this coordinate transformation relation is set up, step 103 can, further according to this coordinate transformation relation, determine a block of pixels to be corrected of block of pixels after proofreading and correct corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected.Specifically, because this coordinate transformation relation can obtain, in the rear block of pixels of this correction, (each proofread and correct the relevant coordinate of rear pixel in this image to be corrected, this second coordinate figure), therefore, if need calculate after this correction in block of pixels the pixel value of pixel after all corrections, just must utilize relevant coordinate, after deciding this correction, block of pixels is relevant to which image information partly in this image to be corrected, that is, determine the rear block of pixels of this correction represents the picture reduction result of which part in this image to be corrected.Therefore, can, by corresponding second coordinate figure of boundary pixel of block of pixels after this correction, decide the image information of its part that institute's wish is proofreaied and correct in this image to be corrected.Detail with reference Fig. 2 about this step, this coordinate transformation relation that after proofreading and correct, the boundary pixel (pixel that dotted line frame 201 comprises) of block of pixels A can be set up by step 101 corresponds to a plurality of coordinate points in solid box 202 in image to be corrected, and the region that solid box 202 is surrounded (, block of pixels A ' to be corrected) image information in, is the image information that after proofreading and correct, block of pixels A institute wish is proofreaied and correct.Then, after determining block of pixels A ' to be corrected, in order conveniently to treat correction pixels block A ', carry out data buffering (buffering), the further geosphere of meeting of the present invention is elected to and comprises less the rectangular-shaped pending block of pixels A of having of block of pixels A ' to be corrected ".And pending block of pixels A " in the pixel value of pending pixel will be the necessary information of the pixel value of each pixel of block of pixels A after interpolative operation goes out to proofread and correct.
In step 105, as pending block of pixels A " after the pixel that comprises determined; meeting of the present invention is further according to pending block of pixels A " in the Rankine-Hugoniot relations of a plurality of pending pixels, the pixel value of each pending pixel is temporary in to a storer (as: static RAM (static random access memory with a plurality of memory banks (memory bank), SRAM)) in, thereby wait for the utilization of follow-up interpolative operation.This step has been used storer configuration skill provided by the present invention, and its fundamental purpose is that the pixel value that accelerates to store is read out the speed of carrying out interpolative operation.For instance, the pixel P who proofreaies and correct in rear block of pixels A of take is example, its second coordinate position in image to be corrected is a P ', therefore, when the pixel value of interpolative operation at calculating pixel P, must use the pixel value of a plurality of specific pending pixel that is adjacent to a P ' (4*4) that dotted line frame 203 (that is, pending block of pixels A ") comprises to carry out interpolation.According to this, the concept of storer configuration skill of the present invention is in a storer read cycle, the 4*4 of an all the plurality of specific pending pixels pixel value is read out, therefore step 105 can be according to pending block of pixels A " in the Rankine-Hugoniot relations of pending pixel all pixel values are deposited in storer, and the plurality of pixel value of a plurality of specific pending pixels is kept in to different memory banks.
The explanation of the skill configuring about sort memory is please further refer to Fig. 2 and Fig. 3.Take Fig. 2 as example, because required information of interpolative operation is the pixel value of 4*4 pending pixel, therefore the pixel value of the pending pixel of the 4*4 in dotted line frame 203 (pending block of pixels A ") can be stored in respectively in Fig. 3 according to the row at its place, in the memory bank A～D of storer 300.For instance, the pixel value of the pending pixel of the first row 2031 in the pending pixel of the 4*4 in dotted line frame 203 (pending block of pixels A ") can be stored in memory bank A; the pixel value of the pending pixel of secondary series 2032 can be stored in memory bank B, and the pixel value of the pending pixel of the 3rd row 2033 can be stored in memory bank C and the pixel value of the pending pixel of the 4th row 2034 can be stored in memory bank D.Please note, above configuration mode is only for explanation is used, and unrestricted, broadly, being characterized as of storer configuration skill of the present invention: (1) is in this storer, for carrying out a plurality of pixel values of the required a plurality of specific pending pixels of interpolative operation, the pixel value of the pending pixel of different lines can be temporary in respectively in the different memory storehouse in this storer, in other words, with regard to a pending block (pending block of pixels A "), there is not the pixel value of the pixel of different lines to be temporary in the different lines address (row address) of same memory bank and (2) if be m*n for the number of the required the plurality of specific pending pixel of interpolative operation, the number of the memory bank of this storer is m, wherein m and n are positive integer.Be this, after by storer of the present invention configuration skill, for the pixel value of the required a plurality of specific pending pixels of interpolative operation, can be read out at a storer read cycle, and then promote the processing speed of interpolative operation.Finally, step 107 can be proofreaied and correct rear pixel for each, first utilize corresponding this second coordinate figure of pixel (as pixel P) after this correction (as a P " coordinate figure) to determine the corresponding a plurality of specific pending pixels of pixel (as pixel P) (4*4 the pixel that dotted line frame 203 is comprised) after this corrections, and a plurality of pixel values that this storer reads the plurality of specific pending pixel certainly carry out an interpolative operation to produce this correction pixel value of pixel afterwards.Although enumerated part actual numerical value in note that above, describe, in the present invention, the number of pixels that interpolative operation utilizes is restriction to some extent not, therefore consider down in difference, the number of pixels that in the present invention, interpolative operation adopts may be different.
Above step 101～107 have represented the flow process of a calculated for pixel values operation of the present invention, in this flow process, can produce after a correction in block of pixels the pixel value of pixel after all corrections by interpolative operation repeatedly, therefore after repeatedly calculated for pixel values operation is performed, can produce before correction image after the correction of the geometric distortion that tool is less.
Because the image to be corrected capturing by fish eye lens or bugeye lens includes image information quite on a large scale, and these image informations are projected onto on a relatively narrow and small image sensing region, therefore cause in original image to be corrected, to have quite manifold lines and pushed and distortion, have quite significantly geometric distortion.Therefore the present invention uses the skill of flake distortion correction to reduce in the lines of distortion in image to be corrected.Moreover, when by fish eye lens or bugeye lens acquisition image, because the scope of acquisition scenery is excessive, make scenery not be parallel to the residing plane in image sensing region completely, thereby make the image capturing often also be accompanied by perspective distortion (perspective distortion).For solving the problem of perspective distortion, the skill that the present invention also proofreaies and correct by perspective distortion is reduced image to be corrected.Therefore, image processing method of the present invention at least needs to carry out flake distortion correction (fisheye distortion correction) and proofreaies and correct (perspective distortion correction) with perspective distortion, so the coordinate transformation relation in the present invention is at least determined by flake distortion correction coordinate conversion and the conversion of perspective distortion calibration coordinate.The most in broad terms, the first coordinate figure (Xc of pixel after aforesaid correction, Yc) can change to be converted to the second coordinate figure (Xd by flake distortion correction coordinate conversion and perspective distortion calibration coordinate, Yd), yet, the demand of adjusting for picture geometrically in order more to press close to user, in other embodiments of the invention, coordinate transformation relation is also further by proportional zoom (scaling) coordinate conversion, area zoom (zoom) coordinate conversion, perspective compensation (perspective compensation) coordinate conversion, perpendicular line and horizontal line compensation (vertical and horizontal line compensation) coordinate conversion and rotation (rotation) coordinate conversion determine, and parameter wherein can be adjusted according to visual perception voluntarily by user.For instance, if it is large not that user feels to proofread and correct the shown object ratio of rear picture, user can select to adjust the zooming parameter value (scaling ratio) of suitable vertical and horizontal, and these parameters can be used to carry out proportional zoom coordinate conversion, and then integral image coordinate transformation relation, change the size of proofreading and correct rear picture.Moreover use can also can be decided in its sole discretion, the part of institute's wish reduction in image to be corrected, for instance, take Fig. 4 image to be corrected and proofread and correct after image be example, can be as seen from Figure 4, the view data that after proofreading and correct, image comprises is only a part for original image to be corrected.About this operation, can in image to be corrected, select voluntarily a reference center by user, and the position of this reference center can be referenced when setting up this coordinate transformation relation, after this reference center is determined, the image only being stretched out in a particular range by this reference center can be reduced, therefore when set up this coordinate transformation relation, proofread and correct pixel after the correction of outermost in image afterwards, only correspond to the boundary position of this particular range.Therefore along with selecting of different reference centers, user can select the part of institute's wish reduction in image to be corrected.
Fig. 5 is the relation having illustrated about between different coordinate conversion and coordinate figure, in view of the details of this coordinate conversion is that the people who has the knack of the technical field of the invention knows, therefore seldom repeat at this.
In addition, the present invention's parameter that conversion adopts for perspective distortion calibration coordinate especially, proposes a comparatively simple method of estimation, the usefulness that can promote image correction process.First, please refer to Fig. 6, it has illustrated the formation reason of perspective distortion simply.Substantially, the initiation reason of perspective distortion is between object plane (Object plane) OP and picture plane (Image plane) IP also non-parallel, owing to thering is an angle Ψ between the two, so cause upper two sections of not isometric line segment R1 and R2 as object plane OP, scioptics 600 can form picture R1' and the R2 ' with equal length while being incident upon as planar I P above.Therefore the imaging on picture planar I P and the material object on object plane be drop to some extent, causes perspective distortion.Be this, if need corrected perspective distortion, must understand the actual numerical value of angle Ψ.Yet, for fish eye lens or bugeye lens, because the characteristic of this type camera lens and general camera lens greatly differ from each other, therefore the estimation mode of angle Ψ is very complicated.Therefore, the invention provides the numerical value that a kind of estimation mode comparatively easily obtains angle Ψ.
Detailed description about above-mentioned estimation mode please refer to Fig. 7.As shown in the figure, object plane OP and as there being angle Ψ between planar I P.The incident ray LN that stems from 1 X on object plane and penetrate photocentre (Optical center) OC of fish eye lens 600 can image in the position of the upper X ' of picture planar I P in theory, yet, if while supposing that incidence angle θ that incident ray LN penetrates photocentre OC equals the refraction angle θ of refracted ray LN', refracted ray LN' and a virtual curved face FLR meeting thereby intersect at intersection point X ".Therefore, from figure, being observed, with the some X on virtual curved face FLR " a tangent virtual representation planar I P' can be same as angle Ψ and incidence angle θ with the angle Ψ ' as plane originally.In brief, the present invention compensates for the hypothesis of simplifying the relation at fish-eye incident angle and refraction angle, therefore can utilize easily the phenomenon of Ψ=θ to decide perspective distortion to proofread and correct required angle parameter Ψ by virtual curved face FLR.In addition, user can be with visual impression, as: to proofreading and correct the perception of rear image, also further readjust by a small margin voluntarily the estimated angle parameter Ψ going out of above-mentioned method of estimation, promote the precision of angle parameter Ψ.
Embodiments of the invention also provide a kind of method for correcting image based on above designed image correction system, please refer to the function block schematic diagram of the image correction system 800 shown in Fig. 8.As shown in the figure, image correction system 800 includes a coordinate transformation relation generation circuit 801, a block selecting circuit 802, a storer 803 and an interpolative operation circuit 804.Image correction system 800 in order to the image Pic to be corrected according in an image source 810 to export image Pic ' after a correction, wherein, image source 810 may be for by fish eye lens or captured real-time (real-time) image of bugeye lens, or is stored in the still image in a memory storage.Moreover image correction system 800 is in order to carry out repeatedly calculated for pixel values operation, each calculated for pixel values operates the pixel value of proofreading and correct the rear pixel of each correction in rear block of pixels in order to one in image Pic ' after generation correction.Moreover, coordinate transformation relation produces circuit 801 in order to set up image Pic to be corrected and to proofread and correct the coordinate transformation relation between rear image Pic ', make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected, and coordinate transformation relation generation circuit 801 can for example, decide this coordinate transformation relation with reference to picture dimension and the coordinate transformation parameter (: reference center, angle parameter Ψ) of image Pic ' after proofreading and correct.Block selecting circuit 802 is coupled to coordinate transformation relation and produces circuit 801, and in order to according to this coordinate transformation relation, chooses a pending block of pixels that at least comprises this block of pixels to be corrected in image Pic to be corrected.Storer 803 is coupled to block selecting circuit 802, there are a plurality of memory banks, and in order to the pixel value of each pending pixel is temporary in respectively in the plurality of memory bank, wherein this pending block of pixels has the pending pixel of multiple row (row), and the pixel value of the pending pixel of different lines can be temporary in respectively in the different memory storehouse in storer 803.Storer 803 may be a static RAM (static random access memory, SRAM).Moreover, interpolative operation circuit 804 is coupled to storer 803 and produces circuit 801 with coordinate transformation relation, and interpolative operation unit 804 utilizes after this correction corresponding this second coordinate figure of pixel to determine the corresponding a plurality of specific pending pixels of pixel after this corrections, and carry out an interpolative operation to produce this correction pixel value of pixel afterwards from a plurality of pixel values that storer 803 reads the plurality of specific pending pixel.So, by image correction system, carry out calculated for pixel values for more than 800 time and operate, can obtain the pixel value of proofreading and correct all pixels in rear image Pic ', and then image Pic ' after output calibration.Because image correction system 800 is that method for correcting image based on above-mentioned is designed, therefore, detailed functions and principle of operation about each circuit in image correction system 800 connect similar in appearance to method for correcting image described above, please reader voluntarily with reference to the above-mentioned explanation paragraph for method for correcting image, at this, seldom repeat.
Below only for coordinate transformation relation, producing circuit 801 describes.Due to, image rectification of the present invention at least needs to carry out flake distortion correction (fisheye correction) and proofreaies and correct (perspective correction) with perspective distortion, therefore producing circuit 801, the coordinate transformation relation in the present invention at least comprises flake distortion correction coordinate transformation circuit and perspective distortion calibration coordinate change-over circuit (not shown), therefore the most in broad terms, coordinate transformation relation produces this coordinate transformation relation that circuit 801 determines at least flake distortion correction coordinate transformation circuit and perspective distortion calibration coordinate change-over circuit determines.Yet, the demand of adjusting for picture geometrically in order more to press close to user, in other embodiment of image correction system of the present invention, coordinate transformation relation produces circuit 801 and also further comprises proportional zoom coordinate transformation circuit, area zoom coordinate transformation circuit, perspective compensation coordinate transformation circuit, perpendicular line and horizontal line compensation coordinate transformation circuit and rotational coordinates change-over circuit.Relation between these circuit and first, second coordinate figure can be with reference to figure 5.Should be noted that, kind and the quantity of coordinate transformation circuit that position fixing transformational relation produces the inside of circuit of not fighting to the finish image correction system of the present invention limits to some extent, in other words, under the prerequisite without prejudice to main concept of the present invention, coordinate transformation relation of the present invention produces circuit may include the kind more than the coordinate conversion shown in Fig. 5, or be less than the kind of the coordinate conversion shown in Fig. 5, and the adjustment geometrically that user can carry out also can therefore and to some extent change.
Generally speaking, the present invention is directed to the image that fish eye lens or bugeye lens capture, a kind of image correction system and method that has good computing velocity and calibration result is provided, and is available for users to adjust image after the correction of output with personal like, and then obtain visual best impression.
The foregoing is only the preferred embodiments of the present invention, all equalizations of doing according to the claims in the present invention book change and modify, and all should belong to covering scope of the present invention.
Claims (10)
1. a method for correcting image, in order to according to an image to be corrected to export image after a correction, wherein image includes block of pixels after a plurality of corrections after this correction, this method for correcting image carries out repeatedly calculated for pixel values operation, the operation of each calculated for pixel values is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct, and includes:
Set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected;
According to a coordinate transformation relation, decision is proofreaied and correct a block of pixels to be corrected of rear block of pixels corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected, wherein this pending block of pixels has the pending pixel of multiple row;
The pixel value of each pending pixel is kept in respectively to having in a storer of a plurality of memory banks, and wherein the pixel value of the pending pixel of different lines can be temporary in respectively in the different memory storehouse in this storer; And
Utilize after this correction corresponding this second coordinate figure of pixel to determine the corresponding a plurality of specific pending pixels of pixel after this corrections, and a plurality of pixel values that this storer reads the plurality of specific pending pixel certainly carry out an interpolative operation to produce this correction pixel value of pixel afterwards;
Wherein the plurality of pixel value of the plurality of specific pending pixel is kept in to different memory banks.
2. method for correcting image as claimed in claim 1, wherein the number of the plurality of specific pending pixel is m*n, the number of the plurality of memory bank of this storer is m, and m and n are positive integer.
3. method for correcting image as claimed in claim 1, wherein this image to be corrected captures by a fish eye lens.
4. method for correcting image as claimed in claim 3, wherein this coordinate transformation relation is at least determined by a perspective distortion calibration coordinate conversion institute; And in the conversion of this perspective distortion calibration coordinate, corresponding to an object plane and an angle parameter looking like between plane, with an incident angle of the corresponding incident ray of this image to be corrected, decide, its this incident light linear system penetrates this fish-eye photocentre.
5. a method for correcting image, in order to according to an image to be corrected to export image after a correction, wherein image includes block of pixels after a plurality of corrections after this correction, and this image to be corrected captures by a fish eye lens, this method for correcting image carries out repeatedly calculated for pixel values operation, the operation of each calculated for pixel values is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct, and includes:
Set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected, wherein, this coordinate transformation relation is at least determined by a perspective distortion calibration coordinate conversion, and in this perspective distortion calibration coordinate conversion, corresponding to the angle parameter between an object plane and a picture plane, an incident angle with the corresponding incident ray of this image to be corrected decides, and this incident light linear system penetrates this fish-eye photocentre,
According to this coordinate transformation relation, determine a block of pixels to be corrected of block of pixels after proofreading and correct corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected; And
Utilize rear corresponding this second coordinate figure of pixel of this correction to determine the rear corresponding a plurality of specific pending pixels of pixel of this correction, and utilize a plurality of pixel values of the plurality of specific pending pixel to carry out an interpolative operation to produce a pixel value of the rear pixel of this correction.
6. an image correction system, in order to according to an image to be corrected to export image after a correction, wherein image includes block of pixels after a plurality of corrections after this correction, this image correction system carries out repeatedly calculated for pixel values operation, the operation of each calculated for pixel values is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct, and this image correction system includes:
One coordinate transformation relation produces circuit, in order to set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected;
One block selecting circuit, be coupled to this coordinate transformation relation and produce circuit, in order to according to this coordinate transformation relation, decision is proofreaied and correct a block of pixels to be corrected of rear block of pixels corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected, wherein this pending block of pixels has the pending pixel of multiple row;
One storer, is coupled to this block selecting circuit, has a plurality of memory banks, in order to the pixel value of each pending pixel is temporary in respectively in the plurality of memory bank; And
One interpolative operation circuit, be coupled to this storer and this coordinate transformation relation and produce circuit, corresponding this second coordinate figure of pixel to be to determine the corresponding a plurality of specific pending pixels of pixel after this corrections after this this correction of interpolative operation unit by using, and a plurality of pixel values that this storer reads the plurality of specific pending pixel certainly carry out an interpolative operation to produce this correction pixel value of pixel afterwards;
Wherein, the pixel value of the pending pixel of different lines can be temporary in respectively in the different memory storehouse in this storer.
7. image correction system as claimed in claim 6, wherein the number of the plurality of specific pending pixel is m*n, the number of the plurality of memory bank of this storer is m, and m and n are positive integer.
8. image correction system as claimed in claim 6, wherein this image to be corrected captures by a fish eye lens.
9. image correction system as claimed in claim 8, wherein this coordinate transformation relation generation circuit comprises at least one perspective distortion calibration coordinate change-over circuit, and this coordinate transformation relation is at least determined by this perspective distortion calibration coordinate change-over circuit, and in this perspective distortion calibration coordinate change-over circuit, the angle parameter between plane corresponding to an object plane and a picture that it adopts, an incident angle with the corresponding incident ray of this image to be corrected decides, and this incident light linear system penetrates this fish-eye photocentre.
10. an image correction system, in order to according to an image to be corrected to export image after a correction, wherein image includes block of pixels after a plurality of corrections after this correction, and this image to be corrected captures by a fish eye lens, this image correction system carries out repeatedly calculated for pixel values operation, the operation of each calculated for pixel values is in order to the pixel value of pixel after producing after a correction in block of pixels each and proofreading and correct, and this image correction system includes:
One coordinate transformation relation produces circuit, in order to set up the coordinate transformation relation between the rear image of this image to be corrected and this correction, make one first coordinate figure of the rear pixel of each correction of the rear block of pixels of this correction correspond to one second coordinate figure in this image to be corrected, wherein, this coordinate transformation relation produces circuit and comprises at least one perspective distortion calibration coordinate change-over circuit, and this coordinate transformation relation is at least determined by this perspective distortion calibration coordinate change-over circuit, and the angle parameter between plane corresponding to an object plane and a picture using in this perspective distortion calibration coordinate change-over circuit, an incident angle with the corresponding incident ray of this image to be corrected decides, and this incident light linear system penetrates this fish-eye photocentre,
One block selecting circuit, be coupled to this coordinate transformation relation and produce circuit, in order to according to this coordinate transformation relation, determine a block of pixels to be corrected of block of pixels after proofreading and correct corresponding to this, certainly to choose a pending block of pixels that at least comprises this block of pixels to be corrected in this image to be corrected; And
One interpolative operation circuit, be coupled to storer and this coordinate transformation relation and produce circuit, corresponding this second coordinate figure of pixel to be to determine the corresponding a plurality of specific pending pixels of pixel after this corrections after this this correction of interpolative operation unit by using, and utilizes a plurality of pixel values of the plurality of specific pending pixel to carry out an interpolative operation to produce this correction pixel value of pixel afterwards.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010558117.3A CN102479379B (en) | 2010-11-19 | 2010-11-19 | Image rectification method and relevant image rectification system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010558117.3A CN102479379B (en) | 2010-11-19 | 2010-11-19 | Image rectification method and relevant image rectification system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102479379A CN102479379A (en) | 2012-05-30 |
| CN102479379B true CN102479379B (en) | 2014-09-03 |
Family
ID=46092012
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010558117.3A Active CN102479379B (en) | 2010-11-19 | 2010-11-19 | Image rectification method and relevant image rectification system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102479379B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107516292A (en) * | 2017-08-30 | 2017-12-26 | 广州视源电子科技股份有限公司 | Method for correcting image, device, equipment and computer-readable recording medium |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295370B (en) * | 2012-09-13 | 2015-07-22 | 上海凯融信息科技有限公司 | Method and system for preventing myopia by monitoring distance between eyes and screen |
| TWI578271B (en) * | 2012-10-23 | 2017-04-11 | 義晶科技股份有限公司 | Dynamic image processing method and dynamic image processing system |
| TWI505203B (en) * | 2012-11-02 | 2015-10-21 | Avisonic Technology Corp | Image processing method and image processing apparatus for generating vehicular image |
| CN104135608B (en) * | 2014-05-21 | 2017-07-14 | 张家港恩达通讯科技有限公司 | The flattening method and system of a kind of fisheye image |
| CN105608712B (en) * | 2015-12-11 | 2018-10-26 | 中国航空工业集团公司西安航空计算技术研究所 | A kind of 3D graphics process perspective correction processing method |
| CN106570842B (en) * | 2016-11-10 | 2019-11-15 | 浙江大华技术股份有限公司 | A kind of antidote and device of fish eye images |
| WO2018102990A1 (en) | 2016-12-06 | 2018-06-14 | SZ DJI Technology Co., Ltd. | System and method for rectifying a wide-angle image |
| CN108307089B (en) * | 2017-01-13 | 2020-06-12 | 联发科技股份有限公司 | Distorted picture correcting device and method |
| TWI602154B (en) * | 2017-04-26 | 2017-10-11 | 偉詮電子股份有限公司 | Panoramic image stitching method and system thereof |
| TWI824321B (en) * | 2021-07-02 | 2023-12-01 | 創惟科技股份有限公司 | Image controller, image processing system and image modifying method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1963838A (en) * | 2006-11-07 | 2007-05-16 | 中国物品编码中心 | Method for correcting distortion of two-dimension bar figure |
| CN101685537A (en) * | 2008-09-25 | 2010-03-31 | 八方科技股份有限公司 | Method for enhancing graphics by using area gain calibration |
| CN101697105A (en) * | 2009-10-26 | 2010-04-21 | 广东威创视讯科技股份有限公司 | Camera type touch detection positioning method and camera type touch detection system |
| CN101742125A (en) * | 2008-11-27 | 2010-06-16 | 义晶科技股份有限公司 | Image processing method and associated device for correcting fisheye image and reducing perspective distortion |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101031052B (en) * | 2006-02-28 | 2010-11-03 | 上海奇码数字信息有限公司 | Method for mapping image address in memory |
| TW200939135A (en) * | 2008-03-04 | 2009-09-16 | Avisonic Technology Corp | Apparatus and method of image calculating process and face detection system using the same |
-
2010
- 2010-11-19 CN CN201010558117.3A patent/CN102479379B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1963838A (en) * | 2006-11-07 | 2007-05-16 | 中国物品编码中心 | Method for correcting distortion of two-dimension bar figure |
| CN101685537A (en) * | 2008-09-25 | 2010-03-31 | 八方科技股份有限公司 | Method for enhancing graphics by using area gain calibration |
| CN101742125A (en) * | 2008-11-27 | 2010-06-16 | 义晶科技股份有限公司 | Image processing method and associated device for correcting fisheye image and reducing perspective distortion |
| CN101697105A (en) * | 2009-10-26 | 2010-04-21 | 广东威创视讯科技股份有限公司 | Camera type touch detection positioning method and camera type touch detection system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107516292A (en) * | 2017-08-30 | 2017-12-26 | 广州视源电子科技股份有限公司 | Method for correcting image, device, equipment and computer-readable recording medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102479379A (en) | 2012-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102479379B (en) | Image rectification method and relevant image rectification system | |
| JP5437311B2 (en) | Image correction method, image correction system, angle estimation method, and angle estimation device | |
| JP4077869B2 (en) | Light source estimation device, light source estimation system, light source estimation method, image resolution increasing device, and image resolution increasing method | |
| CN103500435B (en) | A kind of adapting to image interpolation method of edge guiding and VLSI implement device thereof | |
| CN107566688B (en) | Convolutional neural network-based video anti-shake method and device and image alignment device | |
| CN105740876B (en) | A kind of image pre-processing method and device | |
| US20200394770A1 (en) | Method for optimal body or face protection with adaptive dewarping based on context segmentation layers | |
| Im et al. | High quality structure from small motion for rolling shutter cameras | |
| CN102905147A (en) | Three-dimensional image correction method and apparatus | |
| CN102156969A (en) | Processing method for correcting deviation of image | |
| JP2007013430A (en) | Moving vector detecting device and method | |
| CN110298282A (en) | Document image processing method, storage medium and calculating equipment | |
| EP2864959A2 (en) | Document unbending and recoloring systems and methods | |
| CN110662014B (en) | Light field camera four-dimensional data large depth-of-field three-dimensional display method | |
| CN111507333A (en) | Image correction method and device, electronic equipment and storage medium | |
| CN111866523B (en) | Panoramic video synthesis method and device, electronic equipment and computer storage medium | |
| CN103902730A (en) | Thumbnail generation method and system | |
| US8223229B2 (en) | Lens shading correction for autofocus and zoom lenses | |
| CN116152121B (en) | Curved surface screen generating method and correcting method based on distortion parameters | |
| Sun et al. | Rolling shutter distortion removal based on curve interpolation | |
| CN111767752A (en) | Two-dimensional code identification method and device | |
| CN101729739A (en) | Method for rectifying deviation of image | |
| CN116403226A (en) | Unconstrained fold document image correction method, system, equipment and storage medium | |
| CN114494034A (en) | Image distortion correction method, device and equipment | |
| CN102469249A (en) | Image correction method and image correction device |
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 |