CN101739679A - System and method for generating three-dimensional depth message - Google Patents
System and method for generating three-dimensional depth message Download PDFInfo
- Publication number
- CN101739679A CN101739679A CN200810178923A CN200810178923A CN101739679A CN 101739679 A CN101739679 A CN 101739679A CN 200810178923 A CN200810178923 A CN 200810178923A CN 200810178923 A CN200810178923 A CN 200810178923A CN 101739679 A CN101739679 A CN 101739679A
- Authority
- CN
- China
- Prior art keywords
- depth information
- dimensional
- border
- dimensional depth
- order
- 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.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a system and a method for generating a three-dimensional depth message. The system for generating the three-dimensional depth message comprises a vanishing point decision device, a classification device and a depth allocation unit, wherein the vanishing point decision device is used for deciding a vanishing point in a two-dimensional image; the classification device is used for classifying a plurality of classification structures; and the depth allocation unit is used for allocating the depth message to each the classification structures. Compared with the conventional method for generating the three-dimensional depth message in the prior art, the embodiment of the invention can reappear faithfully and exactly or show similar three-dimension.
Description
Technical field
The present invention relates to the generation of a kind of three-dimensional depth (3D depth) information, particularly relate to a kind of about detecting vanishing line (vanishing line) to produce the system and method for three-dimensional depth information.
Background technology
When three-dimensional body by camera or video camera and projection mapping during to two dimensional image plane because this kind projection is to be not exclusive many-one conversion, therefore can lose three-dimensional depth information.In other words, can't decide its degree of depth by the picture point after the projection.In order to obtain a complete reproduction or approximate three-dimensional performance, must recover or produce these three-dimensional depth informations, in order to carry out the demonstration of the synthetic or image of figure image intensifying (enhancement), image restoration (restoration), image.
The generation of conventional stereo depth information is the end point (vanishing point) of vanishing line in the detected image (vanishing line) institute polymerization, is as the criterion with this end point again, more gives bigger depth value near end point person.In other words, three-dimensional depth information is assigned in gradient (gradient) mode.The shortcoming of the method is that it fails effectively to consider the prior knowledge (prior knowledge) of various zoness of different in the image.Therefore, have same distance with end point but be positioned at the picture point of zones of different, the method will give identical depth value monotonously.
The production method of another kind of conventional stereo depth information is to distinguish zones of different according to the size and the color of pixel value, gives different depth values for zones of different again.For example, the zone with big pixel value and color can be assigned with bigger depth value.The shortcoming of the method is that it does not have considerable border information among the consideration human visual system (human visual system).Therefore, for being the picture point of the different degree of depth originally, can give identical depth value mistakenly because having same pixel value and color.
Therefore this shows that above-mentioned classic method fails verily or correctly to produce three-dimensional depth information, need the generation system and method that proposes a kind of three-dimensional depth information badly, verily and correctly to reappear or to be similar to out three-dimensional performance.Its product structure, with use, obviously still have inconvenience and defective, and demand urgently further being improved.In order to solve the problem of above-mentioned existence, relevant manufacturer there's no one who doesn't or isn't seeks solution painstakingly, but do not see always that for a long time suitable design finished by development, and common product and method do not have appropriate structure and method to address the above problem, and this obviously is the problem that the anxious desire of relevant dealer solves.Therefore how to found a kind of generation system and method for new three-dimensional depth information, real one of the current important research and development problem that belongs to, also becoming the current industry utmost point needs improved target.
Summary of the invention
The objective of the invention is to, overcome the defective of the generation system and method existence of existing three-dimensional depth information, and provide a kind of generation system and method for new three-dimensional depth information, technical matters to be solved is it verily and is correctly reappeared or to be similar to out three-dimensional performance, is very suitable for practicality.
The object of the invention to solve the technical problems realizes by the following technical solutions.The generation system of a kind of three-dimensional depth information that proposes according to the present invention, it comprises: an end point determination device, in order to the end point in the decision two dimensional image; One sorter is in order to sort out a plurality of taxonomic structures; And a degree of depth dispatch unit, give each those taxonomic structure in order to assign depth information.
The object of the invention to solve the technical problems also can be applied to the following technical measures to achieve further.
The generation system of aforesaid three-dimensional depth information, wherein said end point determination device comprises: a line detecting unit, in order to detect the vanishing line in this two dimensional image; And a vanishing Point Detection Method unit, its according to this vanishing line that detects to determine this end point.
The generation system of aforesaid three-dimensional depth information, wherein said line detecting unit uses Hough transformation to carry out the detection of vanishing line.
The generation system of aforesaid three-dimensional depth information, wherein said line detecting unit comprises: a border detecting unit, in order to detect the border in this two dimensional image; One gauss low frequency filter is in order to reduce the noise of this detection boundaries; One critical value device, in order to the border of deletion less than a preset critical, and reservation is greater than the border of this preset critical; One aggregation apparatus, in order to the adjacent of this detection boundaries but discrete pixel gather together; And an end points hookup mechanism, link up in order to the end points that will assemble pixel, to form this vanishing line.
The generation system of aforesaid three-dimensional depth information, wherein said sorter comprises: a border feature extraction unit, in order to detect the border of this two dimensional image; And a textural classification unit, it is according to this detection boundaries, and this two dimensional image is divided into a plurality of taxonomic structures.
The generation system of aforesaid three-dimensional depth information, wherein said boundary characteristic acquisition unit uses the Tuscany boundary filter to carry out Boundary Detection.
The generation system of aforesaid three-dimensional depth information, wherein said textural classification unit uses clustering technique to cut apart.
The object of the invention to solve the technical problems also realizes by the following technical solutions.The production method of a kind of three-dimensional depth information that proposes according to the present invention, it may further comprise the steps: the end point in the decision two dimensional image; Sort out a plurality of taxonomic structures; And assign depth information and give respectively this taxonomic structure.
The object of the invention to solve the technical problems also can be applied to the following technical measures to achieve further.
The production method of aforesaid three-dimensional depth information, wherein said end point deciding step comprises: detect the vanishing line in this two dimensional image; And according to the vanishing line of this detection to determine this end point.
The production method of aforesaid three-dimensional depth information, wherein said vanishing line detects step and uses Hough transformation.
The production method of aforesaid three-dimensional depth information, wherein said vanishing line detect step and comprise: detect the border in this two dimensional image; Reduce the noise of this detection boundaries; Deletion is less than the border of a preset critical, and reservation is greater than the border of this preset critical; With the adjacent of this detection boundaries but discrete pixel gather together; And the end points that will assemble pixel links up, to form this vanishing line.
The production method of aforesaid three-dimensional depth information, wherein said classification step comprises: the border of detecting this two dimensional image; And, this two dimensional image is divided into a plurality of taxonomic structures according to this detection boundaries.
The production method of aforesaid three-dimensional depth information, wherein said Boundary Detection step is used the Tuscany boundary filter.
The production method of aforesaid three-dimensional depth information, wherein said classification step is used clustering technique.
The present invention compared with prior art has tangible advantage and beneficial effect.As known from the above, for achieving the above object, the invention provides a kind of generation system and method for three-dimensional depth information.According to the vanishing line in the two dimensional image, to detect end point.According to detection boundaries, with two dimensional image classification and be divided into a plurality of taxonomic structures.Assign depth information respectively and give each these taxonomic structure, thereby can be verily and correctly reappear or be similar to out three-dimensional performance.
By technique scheme, the generation system and method for three-dimensional depth information of the present invention has following advantage and beneficial effect at least: according to the above-mentioned embodiment of the invention, with the described conventional stereo depth information of prior art production method by comparison, the embodiment of the invention can be verily and is correctly reappeared or be similar to out three-dimensional performance.
In sum, the present invention has obvious improvement technically, and has tangible good effect, really is a new and innovative, progressive, practical new design.
Above-mentioned explanation only is the general introduction of technical solution of the present invention, for can clearer understanding technological means of the present invention, and can be implemented according to the content of instructions, and for above-mentioned and other purposes, feature and advantage of the present invention can be become apparent, below especially exemplified by preferred embodiment, and conjunction with figs., be described in detail as follows.
Description of drawings
Fig. 1 is that the three-dimensional depth information of the embodiment of the invention produces system schematic.
Fig. 2 is the process step synoptic diagram of the three-dimensional depth information production method of the embodiment of the invention.
Fig. 3 is that the vanishing line of another embodiment of the present invention detects synoptic diagram.
Fig. 4 A to Fig. 4 E is the synoptic diagram how vanishing line is polymerized to the various examples of end point.
100: three-dimensional depth information produces system
10: input media
11: the line detecting unit
12: the vanishing Point Detection Method unit
13: the boundary characteristic acquisition unit
14: the textural classification unit
15: degree of depth dispatch unit
16: output unit
The process step of 20-26: embodiment
110: Boundary Detection
112: Gauss (Gaussian) low-pass filter
114: critical value
116: assemble neighbor
118: link
Embodiment
Reach technological means and the effect that predetermined goal of the invention is taked for further setting forth the present invention, below in conjunction with accompanying drawing and preferred embodiment, its embodiment of generation system and method, structure, method, step, feature and the effect thereof of the three-dimensional depth information that foundation the present invention is proposed, describe in detail as after.
Fig. 1 is that solid (3D) depth information of the embodiment of the invention produces system 100.For the ease of understanding the present invention, include original image, handle in the illustration image of image, result images also simultaneously attached be shown in graphic in.Fig. 2 shows the process step of the three-dimensional depth information production method of the embodiment of the invention.
Line detecting unit 11 is handled two dimensional image, to detect or to identify the line (step 21), particularly vanishing line (vanishing line) in the image.In this manual, " unit " speech can be in order to represent a circuit, a program or its combination.The relevant auxiliary picture of graphic center line detecting unit 11 demonstrates the vanishing line that is overlapped on the original image.In a preferred embodiment, the detection of vanishing line is to use Hough transformation (Hough transform), and it is to belong to a kind of frequency domain (frequency domain) to handle.Yet, also can use other frequency domain conversion (for example fast fourier transform (FFT)) or spatial domain (spatial domain) to handle.Hough transformation is a kind of feature acquisition (feature extraction) technology, it is exposed in United States Patent (USP) the 3rd, 069, No. 654, be entitled as " device and method (Method and Means for Recognizing Complex Patterns) of composite mode identification ", the invention people is Borrow's Hough (Paul Hough); And be exposed in " utilizing Hough transformation to carry out the detection (Use of the Hough Transformation to Detect Lines and Curvesin Pictures) of image cathetus and curve " and (be published in the Comm.ACM in January, 1972, Vol.15,11-15 page or leaf (January 1972)), the author is Richard Du Da (Richard Duda) and Peter's Hart (PeterHart).Hough transformation is specially adapted to contain in the flaw image of noise (be noise, below all be called noise), to identify straight line or curve.In the present embodiment, Hough transformation can detect or identify the line, particularly vanishing line in the image effectively.
Fig. 3 is that the vanishing line of another embodiment of the present invention detects.In this embodiment, at first carry out border (edge) and detect 110, for example use the Sobel bound test technology.Then, use Gauss (Gaussian) low-pass filter to reduce noise (square frame 112).In ensuing square frame 114, keep border, and delete its coboundary greater than preset critical.Moreover, with adjacent but discrete pixel gather together (square frame 116).In square frame 118, will assemble pixel by the end points (end point) of assembling pixel and make further to link (linking), thereby obtain desired vanishing line.
Next, vanishing Point Detection Method unit 12 is according to the detected vanishing line of line detecting unit 11 (step 22), to determine end point (vanishing point).In general, end point is the point for each detection line or its extension institute phase crosslinking polymerization.The relevant auxiliary picture of vanishing Point Detection Method unit 12 demonstrates the end point that is overlapped on the original image in graphic.Fig. 4 A to Fig. 4 E is the synoptic diagram how vanishing line is polymerized to the various examples of end point.Wherein, the end point of Fig. 4 A is positioned at the left side, and the end point of Fig. 4 B is positioned at the right side, and the end point of Fig. 4 C is positioned at the top, and the end point of Fig. 4 D is positioned at the bottom, and the end point of Fig. 4 E is positioned at inside.
Produce in another (bottom) path of system 100 (Fig. 1) at three-dimensional depth information, two dimensional image is handled via boundary characteristic acquisition unit 13, to detect or to identify border or separatrix (boundary) (step 23) between structure or object.Because line detecting unit 11 and boundary characteristic acquisition unit 13 have some overlapping functions, therefore, these two unit can merge shares the single line/Boundary Detection unit of use.
In a preferred embodiment, the acquisition on border is to use Tuscany boundary filter (Canny edgefilter).The Tuscany boundary filter is a kind of preferable boundary characteristic acquisition or detection algorithm, be by John F Tuscany (John F.Canny) 1986 pattern analysis and machine intelligences that develop and be published in IEEE Trans, by name " a kind of border detection algorithm (the A ComputationalApproach to Edge Detection) " of 8:679-714.The Tuscany boundary filter is specially adapted to contain the border of noise.In the present embodiment, the Tuscany boundary filter can capture boundary characteristic effectively, shown in the relevant auxiliary picture of boundary characteristic acquisition unit 13.
Next, border/separatrix feature that textural classification (structure classification) unit 14 is provided according to boundary characteristic acquisition unit 13 is cut apart entire image (segment) and is become several structures (step 24).Specifically, textural classification unit 14 is the cutting techniques that adopt based on classification (classification-based), thereby makes the object with similar texture (texture) be connected to same structure.Shown in the relevant auxiliary picture of textural classification unit 14 in graphic, its entire image is divided into four structures or fragment: ceiling (ceiling), ground (ground), right vertical (vertical) face, left vertical plane.This kind is not limited to top described person based on the employed pattern of cutting techniques (pattern) of classification.For example, for the image that get the open air, its entire image can be categorized into following structure: sky, ground, vertical plane and surface level.
In a preferred embodiment, can use cluster (clustering) technology (for example K mean algorithm (k-means)) to carry out cutting apart of textural classification unit 14 or classify.At first, according to the brightness histogram (histogram) of image to determine some classes (cluster).Then, determine the distance of each pixel and all kinds of center (centroid or center), make the short-range similar pixel of tool be gathered in same class (cluster), thereby formation is cut apart or taxonomic structure.
Afterwards, degree of depth dispatch unit 15 is assigned depth information to each taxonomic structure (step 25).In general, the method for distribution of the depth information of each taxonomic structure differs from one another, and but, two or more structures also can adopt identical method of distribution.According to prior knowledge (prior knowledge), the depth value that ground is assigned is less than the depth value of ceiling/sky.Degree of depth dispatch unit 15 is as the criterion with end point usually, adopts gradient (gradient) mode to make the pixel of close end point have bigger depth value.
The above, it only is preferred embodiment of the present invention, be not that the present invention is done any pro forma restriction, though the present invention discloses as above with preferred embodiment, yet be not in order to limit the present invention, any those skilled in the art, in not breaking away from the technical solution of the present invention scope, when the method that can utilize above-mentioned announcement and technology contents are made a little change or be modified to the equivalent embodiment of equivalent variations, in every case be the content that does not break away from technical solution of the present invention, according to technical spirit of the present invention to any simple modification that above embodiment did, equivalent variations and modification all still belong in the scope of technical solution of the present invention.
Claims (14)
1. the generation system of a three-dimensional depth information is characterized in that it comprises:
One end point determination device is in order to the end point in the decision two dimensional image;
One sorter is in order to sort out a plurality of taxonomic structures; And
One degree of depth dispatch unit is given each those taxonomic structure in order to assign depth information.
2. the generation system of three-dimensional depth information according to claim 1 is characterized in that wherein said end point determination device comprises:
One line detecting unit is in order to detect the vanishing line in this two dimensional image; And
One vanishing Point Detection Method unit, its according to this vanishing line that detects to determine this end point.
3. the generation system of three-dimensional depth information according to claim 2 is characterized in that wherein said line detecting unit uses Hough transformation to carry out the detection of vanishing line.
4. the generation system of three-dimensional depth information according to claim 2 is characterized in that wherein said line detecting unit comprises:
One border detecting unit is in order to detect the border in this two dimensional image;
One gauss low frequency filter is in order to reduce the noise of this detection boundaries;
One critical value device, in order to the border of deletion less than a preset critical, and reservation is greater than the border of this preset critical;
One aggregation apparatus, in order to the adjacent of this detection boundaries but discrete pixel gather together; And
One end points hookup mechanism links up in order to the end points that will assemble pixel, to form this vanishing line.
5. the generation system of three-dimensional depth information according to claim 1 is characterized in that wherein said sorter comprises:
One border feature extraction unit is in order to detect the border of this two dimensional image; And
One textural classification unit, it is according to this detection boundaries, and this two dimensional image is divided into a plurality of taxonomic structures.
6. the generation system of three-dimensional depth information according to claim 5 is characterized in that wherein said boundary characteristic acquisition unit uses the Tuscany boundary filter to carry out Boundary Detection.
7. the generation system of three-dimensional depth information according to claim 1 is characterized in that wherein said textural classification unit uses clustering technique to cut apart.
8. the production method of a three-dimensional depth information is characterized in that it may further comprise the steps:
End point in the decision two dimensional image;
Sort out a plurality of taxonomic structures; And
Assign depth information and give respectively this taxonomic structure.
9. the production method of three-dimensional depth information according to claim 8 is characterized in that wherein said end point deciding step comprises:
Detect the vanishing line in this two dimensional image; And
According to the vanishing line of this detection to determine this end point.
10. the production method of three-dimensional depth information according to claim 9 is characterized in that wherein said vanishing line detects step and uses Hough transformation.
11. the production method of three-dimensional depth information according to claim 9 is characterized in that wherein said vanishing line detects step and comprises:
Detect the border in this two dimensional image;
Reduce the noise of this detection boundaries;
Deletion is less than the border of a preset critical, and reservation is greater than the border of this preset critical;
With the adjacent of this detection boundaries but discrete pixel gather together; And
The end points of this gathering pixel is linked up, to form this vanishing line.
12. the production method of three-dimensional depth information according to claim 8 is characterized in that wherein said classification step comprises:
Detect the border of this two dimensional image; And
According to this detection boundaries, this two dimensional image is divided into a plurality of taxonomic structures.
13. the production method of three-dimensional depth information according to claim 12 is characterized in that wherein said Boundary Detection step use Tuscany boundary filter.
14. the production method of three-dimensional depth information according to claim 8 is characterized in that wherein said classification step use clustering technique.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810178923A CN101739679A (en) | 2008-11-27 | 2008-11-27 | System and method for generating three-dimensional depth message |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810178923A CN101739679A (en) | 2008-11-27 | 2008-11-27 | System and method for generating three-dimensional depth message |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101739679A true CN101739679A (en) | 2010-06-16 |
Family
ID=42463130
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200810178923A Pending CN101739679A (en) | 2008-11-27 | 2008-11-27 | System and method for generating three-dimensional depth message |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101739679A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034242A (en) * | 2010-12-24 | 2011-04-27 | 清华大学 | Method and device for generating planar image three-dimensional conversion depth for vanishing point detection |
| CN103426144A (en) * | 2012-05-17 | 2013-12-04 | 佳能株式会社 | Method and device for deblurring image having perspective distortion |
-
2008
- 2008-11-27 CN CN200810178923A patent/CN101739679A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034242A (en) * | 2010-12-24 | 2011-04-27 | 清华大学 | Method and device for generating planar image three-dimensional conversion depth for vanishing point detection |
| CN102034242B (en) * | 2010-12-24 | 2013-07-17 | 清华大学 | Method and device for generating planar image three-dimensional conversion depth for vanishing point detection |
| CN103426144A (en) * | 2012-05-17 | 2013-12-04 | 佳能株式会社 | Method and device for deblurring image having perspective distortion |
| CN103426144B (en) * | 2012-05-17 | 2016-05-11 | 佳能株式会社 | For making the method and apparatus of the image deblurring with perspective distortion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2017190574A1 (en) | Fast pedestrian detection method based on aggregation channel features | |
| Felzenszwalb et al. | Efficient graph-based image segmentation | |
| Kulkarni | Color thresholding method for image segmentation of natural images | |
| WO2021051604A1 (en) | Method for identifying text region of osd, and device and storage medium | |
| US20100079453A1 (en) | 3D Depth Generation by Vanishing Line Detection | |
| JP2013196682A (en) | Group-of-people detection method and group-of-people detector | |
| Ni et al. | Automatic detection and counting of circular shaped overlapped objects using circular hough transform and contour detection | |
| CN105044114A (en) | Electrolytic capacitor appearance package defect image detection system and electrolytic capacitor appearance package defect image detection method | |
| CN108985169A (en) | Across the door operation detection method in shop based on deep learning target detection and dynamic background modeling | |
| CN204789357U (en) | Electrolytic capacitor outward appearance packing defect image detection system | |
| CN104268520A (en) | Human motion recognition method based on depth movement trail | |
| CN104200197A (en) | Three-dimensional human body behavior recognition method and device | |
| CN104597057A (en) | Columnar diode surface defect detection device based on machine vision | |
| Felzenszwalb et al. | E ciently computing a good segmentation | |
| Pahwa et al. | Locating 3D object proposals: A depth-based online approach | |
| Jiao et al. | Color image-guided boundary-inconsistent region refinement for stereo matching | |
| Naumann et al. | Refined plane segmentation for cuboid-shaped objects by leveraging edge detection | |
| CN101739679A (en) | System and method for generating three-dimensional depth message | |
| US20150131897A1 (en) | Method and Apparatus for Building Surface Representations of 3D Objects from Stereo Images | |
| Lin et al. | Irregular shapes classification by back-propagation neural networks | |
| CN105930813B (en) | A method of detection composes a piece of writing this under any natural scene | |
| Zhang et al. | Multi-scale continuous gradient local binary pattern for leaky cable fixture detection in high-speed railway tunnel | |
| Song et al. | Edge color distribution transform: an efficient tool for object detection in images | |
| Xu et al. | A multitarget visual attention based algorithm on crack detection of industrial explosives | |
| TW201015491A (en) | 3D depth generation by vanishing line detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20100616 |