TW201117132A - Method for identifying moving foreground objects in an orthorectified photographic image - Google Patents

Abstract

Photographic images recorded with mobile mapping vehicles (20) in real life situations usually contain cars or other moving objects (34) that cover visual information on the road surface (24). According to the techniques of this invention, moving objects (34) are detected by grayscale differencing in overlapping pixels or sections of two or more orthorectified image tiles. Based on moving object identification, masks are generated for each orthorectified tile. The masks are then compared and priorities established based on grayscale values associated with the masks. Mosaics of a large surface of interest such as the Earth can be assembled from a plurality of overlapping photographic images with moving objects (34) largely removed from the resulting mosaic.

Description

201117132 VI. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention relates to a method for recognizing a moving foreground object in a correcting photographic image. [Prior Art] A digital map and a digital map database are used in a navigation system. Digital maps can be obtained by a variety of methods, including high-resolution imaging from space and positive correction images acquired from land-based mobile vehicles. In the latter case, the images obtained from the land-based mapping system must be converted to a scale-corrected image of the ground features as viewed from above the precise ground position of the ground features. The positive correction image is an aerial image that has been geometrically corrected such that the scale of the image is - (meaning the picture can be considered equivalent to a map). A positive correction image can be used to measure the true distance because it is a quasi-4 representation of the surface of interest (e.g., the 'Earth surface'). Adjust the correct image for terrain fluctuations, lens distortion, and camera tilt. The positive correction image can be obtained very efficiently from the aerial image. However, errors are often introduced, which can lead to inaccurate mapping of geolocation data. _ A question ◎ For normal aerial imagery not obtained completely perpendicular to the Earth's surface. Even when the image is acquired close to vertical, only its exact centerline will be vertical. In order to correct this image, additional terrain height information must be obtained. The lack of accurate height information of objects in the image of the I image can be used to determine the correctness of the image by a two-corner measurement method that can cause up to a dozen meters or more of such images. Correctness can be improved by acquiring overlapping images and comparing the same surface obtained from subsequent images. However, the correctness obtained from this method is 144236.doc 201117132 ‘ There is a limit compared to its cost. As used herein, the term "horizontal" data or information corresponds to an object having a surface that is parallel or substantially parallel to the surface of the earth. The term "vertical" data or information corresponds to an object that can be viewed by a general i ± „ basket of ten views on the viewing axis of the Earth's surface. Vertical information cannot be obtained from a typical overhead aerial or satellite imagery. For land vehicles, such as vans or cars, 〇❻ may also be used for aerial vehicles to collect action data for enhanced digital map databases. Action carts usually have many cameras installed, some of which may be Stereo cameras and all of them are accurately positioned due to the accurate GPS and another position and orientation determination device (eg, inertial navigation system _INS). While driving on the road network or established route, Geocoding a sequence of images in a continuous frame or image. Geocoding means the location and/or possible additional heading and/or directional data associated with the image by the GPS receiver and (possibly) INS. A meta-data attached to each image captured by the camera. The action cart records more than one sequence of images of the surface of interest (eg, the road surface) And for each image of the image sequence, accurately determining the geographic location in the geographic coordinate reference system together with the location and orientation data of the image sequence relative to the geographic location. Corresponding. The geographic image sequence is called geography. Encoded image sequences. Other data may also be collected by other sensors, and similarly geocoded. It is known to obtain positive correction tiles for combining the large surface of interest (such as the Earth) Prior art of the Black Eye Mosaic (BEM). An excellent example of this technique 144236.doc 201117132 is described in Applicant's International Publication No. WO/2008/044927, published July 17, 2008. The right to incorporate references may be incorporated by reference and is hereby incorporated by reference in its entirety in its entirety in its entirety the entire disclosures in The quality of the video content. The fact is that 'the images are usually stitched one after the other, and the shingles on the roof overlap one another in sequence. Extremely similar. Although usually effective, it is often the case that moving objects captured in a film image (for example, a motor vehicle passing through a motion system or a motor vehicle passing through a motion charter) appear in the overlay tile. Instead of underlying tiles, the less ideal tiles are overlaid onto the preferred tiles. Therefore, moving foreground objects that partially blur the road map may appear in the completed BEM. Application No. P601 5247 to "Method Of An Apparatus For Pr〇dueing A MulU-Viewpoint Panorama" PCT describes a vertical image sequence used to generate a vertical panorama using a plurality of observation points from a motion cart. method. Use laser scanner data to detect objects close to the camera while producing full scenes. Unsuitable objects in the image are removed by marking portions of the vertical image that should not be used. Then project the part that should be used onto the panoramic surface. The use of laser data (especially in combination with vertical images) is an expensive, cumbersome and less desirable technique for producing positive corrected horizontal images for use in producing bird eye mosaics. Therefore, there is a recognition of moving foreground objects in positive corrected photographic images of the surface of interest that do not rely on the use of laser scanners or other 144236.doc 201117132 cumbersome techniques (especially when existing image data may be available without simultaneous lasers) The need to scan the data). SUMMARY OF THE INVENTION According to the present invention, a method of recognizing a moving foreground object in a photographic image is described in a "-> The method includes the step of providing a first-pre-slice from a first positively corrected image of the surface of interest, the first tile being divided into discrete regions (eg, pixels) and with respect to the table of interest 2 An absolute coordinate position and orientation are associated. From the surface of interest: the second correcting picture provides a second piece that at least partially overlaps the first piece. The second tile is also divided into discrete regions and associated with: an absolute coordinate position and orientation of one of the surfaces of interest. Comparing the second and the second tiles in a uniform region (i.e., ' the region associated with the same absolute coordinate position relative to the closed). A grayscale value of the coincident region in the slice is determined along with the grayscale value in the second tile. Calculate the absolute difference between the grayscale values of these - grayscale values over the - predetermined threshold. Therefore, when comparing the first tile with the second spell =, the grayscale value is very similar, then between the two: small. The difference in grayscale values will be reduced to - the predetermined threshold. The difference between the extreme grayscale values will exceed the _ predetermined threshold. However, the existence of two animals. The value of the sub-private shows a foreground shift. Therefore, the rule of the present invention does not require any laser to correct the movement in the photographic image. The truth is, move 144236.doc 201117132 The detection of an object (for example, a passing car) is based on correcting changes in the horizontal image. [Embodiment] Referring to the drawings, wherein like numerals refer to the same or corresponding parts throughout the plurality of views, the action cart is generally indicated at 20. The action cart 2 is preferably (but not necessarily) a (four) van or car that is fitted with one of the types commonly used in geography applications or the camera 22. The camera 22 is height calibrated to enable the acquired image of the surface 24 of interest (such as a road) to be geocoded with a particular location and orientation. This is typically accomplished by a Gps Receiver % that receives location data from a plurality of satellites 28 orbiting the Earth. In addition, the orientation determination device (e.g., INS) is represented by feature 3A to provide heading data for each image acquired by camera 22. By means of such devices, each of the photographic images acquired by camera 22 is geocoded, meaning that its position as calculated by (4) connector 26 and orientation device 3G, along with (possibly) other heading information, is associated with the image. Union as a meta-data. When the action cart 2 traverses the road surface 24, at the time of the bu? And i + zJ? draws a continuous image of the road surface, where 々 is the time interval between successive images. The impurities are established and small enough to cause successive images of surface 24 to overlap each other at region 32. As shown in Figures 2A-2C, a plurality of cameras 22 can be used in conjunction with the action cart 20 to record a picture of the surface in a wide range and from different angles of view. During the photography of the surface 24 of interest, moving the foreground object 34 (such as the sports car illustrated in Figures 2A-2C) may temporarily obstruct the image of the surface μ relative to each camera 22 at different times. Obfuscated images are particularly annoying when they occur on lane consolidation, intersections, and other related road features. This is due to the importance of these features in mapping applications. Figure 3 "Following the other example of moving the cart 20 to move the foreground object 34. Here, you 丨φ,τ·, _ face-to-face and rear-facing camera 22 at different times • Η *" The overlap region 32 is obstructed by the moving object 34 only in one moment instead of the other. When a plurality of small overlapping photographic images are mosaicted (for example, a face), it is necessary to use the best image of each of the moments at the same time - the region 32 of the surface of interest 24 is photographed. In the case of (as in Fig. 3), the present invention describes a method by which a moving scene object can be recognized and a better quality image is used for generating a mosaic. Figure 4 illustrates a view from the action cartography of Figure 2 as viewed in Figure 3. The trapezoidal dashed line indicates the boundary of the image that is continued by the front facing camera 22 to the surface of interest 2. The foreground moving object 34 is captured in the upper left quadrant of the image ❹ ^ shows the image that has been called after = using one of the techniques described above. The positive correction image is referred to as a tile, and is referred to as a "second" tile 36 in this particular example, but the term is somewhat random. Therefore, for any given, the corresponding received image is placed in the reference system according to the geocoded data to be entered. Corresponding to the ί-edge (the first piece) and the (third piece), the positive correction of the 赘童娅 1 smock image is placed in the same seat =,, 'first middle' so that the overlap between the images can be found

This is depicted graphically. The group T is again specifically referred to Fig. 5'. At this stage, it is not known which of the two pieces of the moving object and the image--the part of the object and the image of interest are 144236.doc 201117132. For the sake of clarity, the term "zone" is used herein to describe a defined portion or region of the entire tile. In practice, one zone will be assigned to each _ = in the digital image, however, reaching the resolution of the fine scale is not always necessary. Figure 6 shows the second panel % of Figure 5 along with the "first" tile 3' overlap portion 32. The first tile 38 represents a photographic image taken by the camera 22 at time μ or immediately prior to the time at which the photographic image of the second tile 36 is being corrected. As will be described later, it is possible that the first tile 第二 and the second 拼 tile 36 are simultaneously acquired by two different cameras 22 or acquired by two different cameras at different times (as prompted in Fig. 3). When the tiles 36, 38 overlap in the manner shown in Figure 6, the non-moving surfaces 24 of interest appear substantially identical, such that the images can be overlapped with little to no distortion. This can be confirmed by the fully aligned lanes in the overlap region 32. However, the moving object 34 has a different position at time & time, and thus can be seen at the overlapping portions of the positively-corrected tiles 36, 38 along the road. The overlapping regions 32 may also be referred to as respective ones (or pixels) of the &32' stomach patches 36, 38 associated with the same absolute coordinate position relative to the surface 24 of interest. Figure 7 and Figure 8 respectively show the consistent area of 筮 # _ # u, 钿, s _ 片 36 and the first splicing piece: 2 2 ie 'Fig. 7 is the fragment view of the second snippet. Lead to section 32. Another ancient door, Figure 8 is a fragment view of the first piece 38, which results in section 32. In the process of comparing the second spelling zone 32 of the first piece, it is apparent that in Fig. 7, the road surface 24' is not obstructed. In the road of Fig. 8, the portion is obstructed by the moving object 34. It is possible to determine whether there is an object in motion by comparing the weight of the patch to the part #44236.doc -10- 201117132 34. This is done by calculating the absolute difference of the grayscale values on a region-by-region or pixel-by-pixel basis. This is then limited to obtain a black/white image called a reticle 40 as depicted in FIG. Whether for pixel-by-pixel or coarser region analysis, the grayscale values are determined across the entire uniform region 32 for each of the first tile 38 and the second tile 36. Grayscale values are typically in the range of 0 to 255, where 〇 is equivalent to black and 255 is equivalent to white. In color images, grayscale values can be calculated by simply averaging individual red, green, and blue color values for each region or pixel. Therefore, according to the simple averaging technique, if the red color value is 丨55, the blue color value is 14 and the green color value is 90, the grayscale color value is about 86. However, in practice, grayscale values are usually calculated as weighted sums. For example: 〇.2989xr+ (K587〇xG+〇.114〇xB. Of course, other grayscale decision techniques can also be used to set the appropriate threshold between the numbers 255 and 255. For example, the second threshold can be selected. In this case, if the absolute value of the absolute 'and the difference between the grayscale values in the pixel or region of the first patch 38 and the third patch, the patch 36 exceeds the threshold value. (e.g., 60), then before moving: Object 34 is identified as being present in the pixel or region. As an example, the grayscale value of a particular pixel or region in the region 32 of the right tile 38 is 86 and the grayscale value in the corresponding pixel or region of the slice 36 is 丨5, The absolute difference between the values then equals 86 minus 15 or 71. The difference 71 is above the exemplary threshold 60' and thus concludes that the moving foreground object 34 is depicted or captured in a particular pixel or region of the coincident region 32. : This way compares two tiles 36, 38, which produces a mask 4 that can be called the 4th (because it is associated with the tile - 38). When the first 144236.doc 201117132 tiles 3 The absolute difference in the grayscale values between the 8 and the second tile 36 is lower than the predetermined threshold value. The first photomask 40 assigns a white grayscale value (i.e., 255) to the first mask. Corresponding pixel or region 'However, when the calculation of the absolute difference produces a number above a predetermined threshold so that the moving foreground object is identified as being present in the pixel or region of the second tile 36, the photomask is Corresponding pixel or region assignment—the black grayscale value (ie, g), represented by the domain. Therefore, in the example mentioned above (where ash (10)= For a difference, the particular pixel or region of the mask 40 will be assigned a black grayscale value or appear black, as shown in Figure 9. By this method, the mask 40 clearly identifies the foreground of the tissue movement. Pixel or region 0 of object 34 Of course, when the absolute difference between two corresponding pixels (or regions) exceeds a threshold, it can be easily reversed by assigning 255 instead of 〇 to a pixel. "Black" convention. A completely different way of explaining this feature of the invention avoids the potentially complicated use of the terms "white" and "black" and instead focuses only on pixel priority or importance. The pixel (or region month) priority can be strictly evaluated based on the grayscale value comparison. In the case of the value setting (in the previous example, the prompt is "60" for the purpose of discussion only) The difference comparison is given a higher priority than the comparison of the comparisons on the opposite side of the margin. Therefore, in a method t 'lower value (also ~ 'below the threshold) means more important pixels And in the other - method 'higher value It is more important pixel. This is just another way to explain the use and implementation of the mask value. Or 'not assign black 〇 to the corresponding pixel or region of material 4G (or white - J2 · 144236.doc 201117132 color 255 Gray scale values, it may be preferable to assign an intermediate gray scale value to a corresponding pixel or region in the reticle, the intermediate gray scale value may be equal to the gray scale value calculated in the coincidence region 32 of the first patch U. In other words, if the corresponding pixel or region in the uniform region 32 of the first tile has a grayscale value of 71, and the calculation of the absolute difference exceeds the predetermined threshold, the corresponding region or pixel in the mask 4〇 will be An intermediate grayscale value of 71 is given. This is an alternative to the method described above and shown in circle 9, such that the reticle 4 will display a threshold (eg, 6 〇) and 0 (or The grayscale value between 255) when the white black convention is reversed as previously described. In any case, the notable system reticle 40 is produced by comparison of two tiles 36, 38, wherein the moving foreground object 34 is calculated by calculating the gray of the corresponding pixel or region in the coincident region 3 2 It is identified by the absolute difference of the order values.

Figure 10 provides an overview of the method steps using a functional module in a practical application, such as a computer processor programmed with an enabling software. A flow of methods for producing a mosaic using a reticle is shown in a simplified manner. According to this technique, the positive corrected image of the road 24 is collected, which has been recorded by the calibrated visual device 22 mounted on the mobile cart 2. Each image victim has a location data corresponding to the correcting tile. A reticle is then created by comparing the overlapping tiles, thereby providing information about the quality of each region or pixel of the region 32 in the positive correction tile. These masks can then be used to create a mosaic of surfaces 24 of great interest, such as the surface of the earth. In this way, a reticle is produced for each positive correction tile by comparing the overlapping positive correction images. However, as described more fully below, the J44236.doc -13 - 201117132 modeling or prediction technique can be used to predict when moving objects 34 will be in a particular tile image and then only produce masks for those tiles. . The detection of moving object 34 can be enhanced or improved by comparing the sequence of reticle, as best shown in Figures 11-13. For example, 圊丨丨 depicts the positive correction of the second tile 36 as shown in FIG. To improve the original detection results, the behavior of the moving object 34 can be modeled. The moving object 34 generally falls into two categories: an object at a substantially constant speed relative to the mobile cart 20, and a moving object 34 that is being chased by the upstream cart 20 or being captured by the cart. Although the object 34 in the first category traveling in front of the action cart 20 does become visible in the top portion of the stitched image, it also disappears from the same portion of the image. Such objects 34 are not difficult to handle in practice because when a continuous tile overlaps the other to produce the resulting mosaic, it is "tile away" and is almost always invisible in the final mosaic. This is because it does not contain a piece of the object below it and is drawn on it, which is very similar to the roof tile. Therefore, objects in the second category (catch up) tend to cause greater difficulty. These objects 34 tend to appear in the resulting mosaic (BEM) or tiles and almost always travel in a different lane than the action cart. This is due to the specific nature of catching up with the car (see Figure 2a for illustration). Figure 12A depicts four areas in the same time (A, B, C) that correct the image or tile at time t. The original detection data of the reticle data in D). Therefore, along the horizontal axis, the numbers W, 2, 3, .. in Fig. 13 indicate the time or frame number with respect to the specific camera 22. According to the figure, the vertical The axis represents the occlusion of the area eight to !) in the horizontal direction. Here, 144236.doc 14 201117132 顶部 Ο in the top part of the image means that an object 34 exists on the left side of the tile (see Figure 11). The black color in the bottom portion of the image means that the object 34 is measured during the first step of the mask generation (Fig. 6), which detects the original movement obstacle %. Therefore, in particular, reference is made to Fig. 12 and In Figure 13, the horizontal frame is divided into four vertical areas Α to D. The area is all black or all white. It is the area with a specific value after the original obstacle is predicted (in eight to..., the value is selected by Z to the total number of pixels (for black 〇 and for white) 255) g) This 'in order to improve robustness, the data is adjusted based on the modeled behavior of the object 34 moving through the frame. The result is the data described in Figure 2β. Figure UB is more clearly depicted over time. In the front (5) solid frame, it quickly catches up with the object 34 of the action cart 2; and then catches up with the much slower moving object μ of the action cart 2 in the frame 约 to about 50. The next 10 frames (approximately) do not contain (d) the measured moving object, however, the frame 7〇 to approximately) shows the action cart 2 chasing the moving object 34.

Each reticle can be described as indicating which of the regions or pixels in the correcting image ((4), the tile contain the data set of the moving object 34. The improvement of the measured data is shown by the foregoing example of Fig. 12A Μ 12Β_ to produce better results. These steps are not performed for each component of the vision system, but are performed with a subset of features. For the subset, the reticle data is easy to produce: Off: ί is available for use. However, based on the detection of the subset of the visual Li: the knowledge of the setting of the visual system of the visual drawing car, it is also possible to correct the image for all the components to generate the photomask and the secret to record the visual system. Different components (including camera 22) do not. The action cart is on the 2nd floor. This means that at times, the object 34 on the road surface 24 144236.doc -15- 201117132 can be seen at different locations in the plurality of vertical images recorded by different components of the vision system. Thus, given knowledge of the position and movement of the object 34 in motion of at least one component 22 of the vision system on the road surface 24, it can be predicted where the moving object 34 on the road surface 24 will be in the image of other components of the vision system. And visible, and can also produce mask data for their components. As an example, a subset of the cameras 22 can be two side cameras (left/right), and the reticle is produced by correcting the difference in the positive correction space for only the two cameras. Based on these results, it is assumed that the moving object 34 conforms to the following assumptions, and a hood can be generated for other cameras (for example, the front camera and the rear camera). For each component of the vision system, if the moving object is at time"

And at time 12, it is visible in the correcting image, and it is expected to be visible for all ^ visible 'S in "< ί < /2 ' and expected to be visible in time " More than one object moves away from the visibility (m〇ve _ 〇f visibiHty) in the opposite part of the image at time G. therefore,

An object 34 that becomes visible on the right side camera 22 produces a hood for the right front camera to enable use of this. Due to the difference in view, the portion of the road 24 that is blocked in the side camera 22 is still visible in the front camera 22, so images from the camera can be used. Once the chasing car becomes visible in the left side of the right camera and the right part becomes unusable again, a mask can be created for the front camera so that the one is not used in this situation (because the obstacle 34 will become more and more visible). Since the heading of each camera 22 and the heading of the camera are known and based only on the angle of the mask of the subset camera, the mask is also produced for other cameras. As long as the common part between the 144236.doc .16- 201117132 that is correcting the space in the space is large enough, it is possible to explicitly create a mask for each camera. However, using only a select subset greatly increases processing speed and only slightly reduces the results. Therefore, the more the obstacle behavior is consistent with the assumptions stated above, the less the reduction in performance noted. As stated above, the reticle can be interpreted as a weighted image. Black (that is, the grayscale value of 255) means the lowest priority, and white means the highest priority. The first two steps in the mask generation method flow only produce black or white values. As previously suggested, the 'third step can produce a grayscale value less than 255' whereby different priorities are given to different cameras based on the reticle of the subset camera and the angle of the camera. By this means, it is possible to optimize the production of the positive correction patches 36, 38 obtained from the vertical image in order to improve the visibility of the road surface and the shoulder. Since the same points on the surface 24 of interest can be seen from two different cameras 22 (or from the same camera 22 at different times) at the same time or at different times, improved visibility can be achieved using the concepts of the present invention.

Another flow chart of the process, wherein the first tile provided by the first positive correction image is read (step 42) together with the first mask-photomask step 44), the identification Any known moving foreground object in a piece of film. For the purposes of the discussion, it can be assumed that the first sister y, the slap- 疋 拼 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同 连同The second piece with the first piece weight ', τ々里登's new positive correction picture, is read by the system its position data H light = up, as indicated at step 48. Again, in step 50, the first piece The reticle of the film is projected onto the temporary reticle - on the frame side, at step 52, the camera distance of the temporary 144236.doc 17 201117132 piece is calculated. The focus of the ECU 6 field measured by the 盍6 field is Considering the pixel or zone-spelt disc in the first phase. The first empty portion coincidence area 32 is compared on a per-regional basis or possibly pixel-by-pixel. If the first- or destination tile has '= then the corresponding region from the second, temporary... Sub-pixels. This is shown in the query 54 and the step 56. The grayscale value of the corresponding pixel or region in the mask is & 1 corresponds to the grayscale value of the pixel or region, and is replaced by _戈=中素素 or region - The image of the piece in the piece is like a pixel or area in the middle of the month. This is shown in the inquiry 58 of the step: . If the grayscale values are equal or at the prompt, then another inquiry is made at 62 to determine whether the second camera is less than the camera distance 56 of the first and destination pixels, and the temporary pixel is closer. For the distance acquisition, the first pixel, the temporary pixel (or region) is copied to (ie, replaced) the first pixel "region". The mask value is then updated (step 64) and the camera distance (in step). An inquiry is made at 68 as to whether the last region or pixel in the consensus zone has been considered. If no, method steps 52 through 66 are repeated. Once the last pixel (or region) from the consistent region is analyzed in this way, the updated tile is saved in step 7A along with the updated mask, and the updated tile and the updated mask become a mosaic Part of (ΒΕΜ). Referring to Figures 15 through 18C, the diagram of Figure 14 is graphically represented. In these examples, the first tile 38 is represented by the forward pointing camera 22 and the second tile % is derived from the angled camera 22. However, it must be understood that the particular orientation of the camera 22 shown in Figure b is strictly for illustrative purposes only. Correction of the first piece - piece 38 is not shown in Figure 16A, and the correcting piece is shown in Figure 17-8. Needle 144236.doc -18- 201117132 Ο 光 The reticle 40 produced for the first piece 38 is shown in Fig. 16A, and the reticle 72 of the second piece 36 is shown in Fig. 17A. In this simplified example, only the moving object 34 (Fig. 17A) is detected in the second tile 36, which corresponds to the reticle 72 reflecting the identified moving object 34 therein. Both the tile and the reticle image are preferably stored in the AVI file. As shown in the figure, there is no object to be obscured in the tile π of the figure, because the moving object has not been detected in the horizontal image. Therefore, the photomask 40 is completely white. The second panel 36 and its reticle 72 are shown in Figures & Figures (10). Next, the tiles 36, 38 are shown in Fig. 18: overlapping without a reticle, such that when the second tile 36 is overlaid with the first tile 38, the object 34 is moved to make the portion of the road image clearly visible in the image ΐ6Α. blurry. Next, the masks 4, 72 are shown as being combined in the diagram (10). By comparing the coincident (ie, overlapping) regions 3 of the first patch 38 and the second patch 36: the gray scale value in the second mask 72 is greater than the gray scale value in the first mask 4 , Bay 1 will replace the first tile % with the consistent region from the second tile 36 - however, in this particular case the opposite case, because the comparison of the corresponding regions of the two optical I commands shows the second The reticle 72 is arranged to have a gray scale value of the corresponding area of the gradation of the photomask 4. Therefore, the underlying portion of the image of the _tile 38 is used, as indicated by the resulting figure (10) jin. Fig. 18C shows the moving object 34_ gram, because the second piece 36 contains 丄 in the resulting Marseille, and there is no shirt in the corresponding area. Therefore, when in the first region, ρ does not exist for the corresponding pixel or the known moving object 34. In the case of (4) rhododendrons in the unanimous zone, the comparison between the eight and the grading of the first reticle and the first hood is substantially equal, and the evaluation will be 144236.doc • 19· 201117132 The distance between each of the first and second images. The distance of the photograph here indicates the distance between the image in the patch and the focus of the camera 22. The image with the smallest photographic distance will be recognized as Reliable, and therefore its image will be given priority in the overlap area U. - Once the overlap is completed, the mosaic mask is updated along with the image distance recorded in the mosaic so that in any subsequent stitching operation, the new correct spelling The slice will be compared to the recorded reticle material. In this way, the positive correction tiles are combined into a mosaic, wherein the overlapping regions are selected based on the image content that is specifically related to the presence of the moving object (four). Thus, by the technique of the present invention, Recognizing the moving object and then producing a reticle from the positive correction patch, which can be used to determine which regions of the overlapping tiles should be prioritized when producing a mosaic of large surfaces 24 of interest, such as the earth. According to the prior art, the positively corrective patch that is over-selectively overlaid can give less useful results because the barrier 34 can cover a portion of the surface of interest 。. However, according to the present invention, the use of a reticle helps the selection to have a level The best available image of the most relevant information on objects such as lanes, lane corridors, drains, etc. Therefore, the use of reticle helps to improve the legibility of the resulting mosaic (BEM). The cover can be produced strictly based on the compared image data, so no additional imaging or laser data techniques are needed to identify the moving object 34. In fact, only a pair of overlapping horizontal (positive corrected) images are needed to produce a bird eye mosaic ( BEM). Detecting moving objects by correcting the gray-scale difference of the common area or pixel of the tiles. Contrary to changing the guess in the vertical frame, because the price measurement system is in the positive correction Therefore, the method can directly distinguish the background from the moving object M. 144236.doc -20· 201117132 FIG. 19 shows two alternative applications of the present invention, in which the positive correction patch is carried on the air. In the case of a vehicle (such as a satellite 12 or an aircraft 22), the image obtained by the phase = 122, 222 is generated. In this case, the a 1 m sand front and the early obstruction 134, 234 can be generated in the resulting image. Obstacle. Through the direct application of the concepts described herein, it is possible to improve the image quality of the resulting mosaic from such aerial photographs. Ο ❹ The foregoing invention has been described in accordance with relevant statutory standards, and thus the description is illustrative in nature. It is to be understood that the embodiments of the present invention will become apparent to those skilled in the art and fall within the scope of the invention. The scope of legal protection conferred by the invention can only be determined by studying the scope of the following patents. Figure 1 is a simplified illustration of the height of a (four) system® vehicle that traverses the road and obtains a series of sequential images using appropriate photographic equipment. (4) Sequence image using GPS positioning data along with directional data geocoding obtained from appropriate telemetry equipment; π” Figure 2A Figure 2 [Description - chronological view in which an actuated cart is driven in accordance with the present invention - a moving object in the foreground (in this case, a sports car) - Figure 3 is an action map showing a moving foreground obstacle a time shifting sequence of the vehicle that partially obscures the image of the desired road surface acquired by a (front facing) camera but does not cause the same surface image acquired by a different (backward facing) camera Blur; The picture is a simplified perspective view of a self-propelled cart (such as the action map 144236.doc •21-201117132 shown in Figure 3, 'and the dotted line indicates the front of the camera from the front to move the camera to get the car) The front-facing camera viewing obstacle appears on the boundary of the photographic image in the front left lane; Figure 5 shows the positive correction view of the photographic image of Figure 4, which shows Obscuring the view of the road for the blackened part of the upper left corner; ', , , and ^ are the pieces of the ride in Figure 5 together with the previous first-heart... (4) which is configured to show the moving foreground obstacles _ a tile to a tile to change the relative position and can be used to create a view obstacle without creating a view obstacle in another tile; Figure 7 depicts the consistency of the second tile as shown in Figure 5. Area; :8 is the view of the consistent area of the first piece of Figure 6. The moving obstacle in the middle of the road does not block one part of the road surface; Figure 94 Tian Hao · used in the uniform area of the first piece (Figure 8) Figure 10 is a flow chart depicting the use of the method of the present invention to create a mosaic; Figure 11 is a schematic representation of a positive correction tile similar to the positive correction tile of Figure 5, which is subdivided for purposes of image improvement after processing In four rows (A to D) Figure 12A is a time diagram of the raw data collected from the present invention, wherein the columns represent the subdivided regions (eight to ten) in each tile and the rows represent the smooth tiles or images. , ί, etc.); Figure 12 is a time diagram of Figure 12, which illustrates the use of behavioral modes. FIG. 13 is an enlarged view of a region limited to 13 in FIG. 12A; FIG. 14 is a view showing a road surface in a positive correction image for improving stitching along the road using a reticle Flow chart of the sequence of steps of visibility; 144236.doc -22- 201117132 Figure 15 is a simplified top view of an action cart equipped with a plurality of cameras, two of which simultaneously photograph the overlapping areas on the surface of interest; Figure 16A A first tile drawn from the first phase before the pointing of the action cart in FIG. 15 is depicted; FIG. 16B is a mask produced for the first tile of FIG. i6A; FIG. 17A is from FIG. The second figure obtained by the action drawing cart facing the second side is correcting the second piece;

Figure 17B shows the second mask produced for the second tile of Figure 17A; Figure 18A shows the stitching of the first tile and the second tile, wherein the overlapping two tiles are visible due to moving the foreground obstacle Part of the road surface is blurred. Figure (10) depicts a comparison between the first mask and the second mask, the mask priority in the basin is evaluated and used to determine which portions of the first and second tiles contain the focus More accurate information on the surface; knife

Figure 18C is a view of the reticle comparison as shown in Figure 18A. Figure 19 is a highly simplified view of a manner in which the concepts of the present invention can be applied, a star image and/or an aerial image. However, Figure 18C shows the use of a borrowed mosaic; and for other image acquisitions and mosaics where the corrective patch can be derived from the [main component symbol description] 13 Area 20 Action Cart 22 Camera 24 Pavement 26 GPS Receiver 144236.doc • 23- 201117132 28 Satellite 30 Orientation device 32 Overlapping area 34 Moving foreground object 36 Second piece 38 First piece 40 Mask 72 Photomask 120 Satellite 122 Camera 134 Moving foreground obstacles 220 Aircraft 222 Camera 234 Moving foreground obstacles A Area B Area C Area D Area H4236.doc -24-

Claims (1)

201117132 VII. Patent application scope 1. 识别 In the method of correcting the photographic image, the method of recognizing the movement is not the body. The method includes the following steps: First correcting the photo from one of the surfaces of interest Providing a first tile, the first tile being divided into discrete regions and associated with an absolute coordinate position and orientation relative to one of the surfaces of interest; from the surface of interest - a second positive corrected image is provided - a second tile at least partially overlapping the first tile, the second tile being divided into discrete regions and associated with an absolute coordinate position and orientation relative to the surface of interest; comparing the first tile a uniform region associated with the same absolute coordinate position of the second tile relative to the surface of interest; and a grayscale value of the uniform region in the first tile of the singer; Calculating the grayscale value of the uniform region in the two tiles, and calculating the absolute difference of the grayscale values between the respective first tiles and the second tiles; and characterized in that The absolute difference of the gray scale values of the consistent region exceeds Identifying a moving foreground object when the threshold is predetermined. 2. The method of claim i, further comprising the steps of: generating a first mask for the first tile, dividing the first mask into corresponding a discrete area of the first patch of the first tile, assigning a white grayscale value to the first when the absolute difference of the grayscale values of the consistent region in the first tile is lower than the predetermined threshold The corresponding area of the reticle, and when the absolute difference of the gray level value of the consistent area in the first tile exceeds the predetermined threshold, the 144236.doc 201117132 is sent to the first reticle Corresponding area. 3. The step of assigning a non-white gray scale value to the square of the request item 2 includes the absolute difference of the gray scale value of the uniform region of the first spell θ τ t exceeding the predetermined threshold When the value is _1, the black grayscale value is assigned to the corresponding area of the first reticle. Early 4 · As in the case of the square 2 of the request item 2, the step of assigning a non-white gray scale value to the eighth is included in The first-slice piece. The absolute difference of the gray-scale value of the far-consistent area of the armor is over-X staff 疋δTM limit The uniform zone value in the first tile is transmitted to the corresponding zone of the first photomask. 5. The method of any one of items 1 to 4, wherein the grayscale value is determined. The k is represented by the 第ί, λ. magic 'work color, green color and blue color value represented by the blending of the respective first-slices and the second piece. 6. For example, the monthly item 1 The method of claim 4, wherein the step of identifying a moving foreground object comprises setting a predetermined threshold between sniffing. 7 · as claimed in any of claims 1 to 4 φ red TS ^ The method wherein the step of identifying a moving foreground object comprises a value of between about 6 255 and 255. 8. The method of any one of clauses 2 to 4, further comprising the steps of: generating a second mask for the second tile, dividing the second mask into the second spell a discrete region of the regions of the slice, assigning a low priority grayscale value to the second photomask when the absolute difference of the grayscale values of the second spelling region is below the predetermined threshold Corresponding zone, and assigning the -high priority grayscale value to the second mask 144236.doc 201117132 when the absolute difference of the grayscale values of the consistent zone in the second tile exceeds the predetermined threshold The corresponding area of the δ hai, and the movement between the ritual child-gu-moving-dawn reticle and the second reticle, the behavior of the object. 9. The method of claim 8, wherein the at least partially riding comprises, in a step T, providing a third associated with a third tile overlapping with a tile, and based on modeling The position of the action of the cattle and the <& ss, al scene object. Predicting a method of moving the first 10.=: item 9, wherein the step of predicting a position of the moving foreground object further comprises: overriding the assigning step with the following steps: even if the second spell The absolute difference of the gray scale values of the coincident region is reduced to the value of the ..., and the high priority gray scale value is still attached to one of the second masks. 11. The method of claim 9, wherein the step of predicting the position of a moving foreground object further comprises overriding the assigning step with the step of: even if the grayscale value of the region in the second tile The absolute difference exceeds the predetermined threshold and a low priority gray scale value is still appended to the first region of the second mask. The method of any one of clauses 1 to 4, wherein the steps of providing the respective first and first tiles comprise installing at least one on a mobile vehicle that moves relative to the surface of interest camera. The method of any one of the preceding claims, wherein the step of associating the first piece with the second piece comprises printing on the respective first piece and the second piece from the first The coordinates of the GPS satellite receiver. 14. The method of any one of the preceding claims, wherein the steps of providing the respective first and second tiles comprise obtaining the first 144236.doc 201117132 photo and at different times The second photo. The method of any one of claims 1 to 4, wherein the steps of providing the respective first and second tiles comprise acquiring the first and second images at the same time . 144236.doc 4-