TW201317803A - Method of uncovering landmarks and its scope by using community website photos - Google Patents

Abstract

A method of uncovering landmarks and its scope by using community website photos, characterized by analyzing and exploring a large number of photos in community websites (such as Facebook and Flickr) or Web Albums (e.g. Picasa), using GPS tags and text labels in the photos and other information uploaded by users, projecting the information onto a map, and then applying spatial low-pass filters to remove scattered or lowly correlated coordinate positions, thus automatically marking the range whereat a landmark is located; by analyzing the characteristics of text labels representation and landmark distribution range, famous landmarks in the city can be further explored. This invention can be applied to an electronic map for automatically labeling the range of a landmark region, it can also be applied in providing tourist advice, automatic photos marking or travel footprint, providing real-time information services such as personal position.

Description

A method for discovering landmarks and their scopes from photos of social networking sites

The present invention relates to a method for discovering landmarks and their scopes from social network photos, in particular for using a web search and data exploration technology to automatically discover famous landmarks and mark them from photos of social websites or web albums. A regionally-wide approach.

In recent years, due to the continuous improvement of digital camera functions and the popularity of the Global Positioning System (GPS), and the increasing bandwidth of the Internet, the number of digital photos shared and uploaded has soared. These digital photos include photographers. GPS tag information at the location, using this information to help improve the performance of image retrieval and bring more geographically relevant application services.

The current technology for marking landmarks is mainly based on the GPS location or density of the user uploading digital photos. A similar concept is mentioned in the prior patented technology: US US 2009/0279794 A1 "Automatic Discovery of Popular Landmarks" The method is based on (1) using the GPS position and visual feature information, first classifying the picture data by the distance (distance) of the position, and then using the visual feature information to classify the second layer in each class, thereby identifying the The landmark is marked with a dot; (2) The number of images uploaded by the user is used to determine whether the landmark is a popular tourist attraction. In addition, there is also a method of using mean shift in the literature to find out where the density of GPS tags is higher in photo data, and to represent the landmarks by points or circles, but how to find the radius of appropriate size to represent this The scope of landmarks is a difficult problem. Moreover, most landmarks are not all circular, for example, riverside parks are almost always strip-shaped. In addition, using dots to represent an area, how many points to use to represent a landmark is another problem.

It can be seen that there are still many shortcomings in the above-mentioned methods of use, which is not a good design, but needs to be improved.

In view of the shortcomings derived from the above-mentioned conventional methods, the inventor of the present invention has improved and innovated, and after painstaking research, he finally succeeded in researching and developing the method for discovering landmarks and their scopes from social network photos.

The purpose of the present invention is that when many famous scenic spots or landmarks are presented on an electronic map, the conventional presentation manner often only points one point, not a regional range, and it is not easy for the user to have a clearer range of the landmarks. concept.

Another object of the present invention is to provide a method for automatically marking landmark area ranges and automatically discovering famous landmarks, which can be applied to map search, providing travel suggestions, automatically labeling photos or travel trails, and providing personal location related instant information.

In order to achieve the above-mentioned invention, a method for discovering landmarks and their scopes from social network photos is to analyze the GPS tags and text label information contained in the uploaded images of the users on the social networking site or the web album, and use the coordinate position conversion. Projecting the coordinate information onto the electronic map, and then using the spatial low-pass filter to remove the scattered or low-correlation position according to its location, the area of the landmark can be marked. In addition, according to the area found by each text label, by analyzing the representativeness of the text label and the distribution characteristics of the landmark, the famous landmarks are explored.

The invention provides a method for automatically excavating landmarks and marking the scope of social network photos. When compared with other conventional technologies, the invention has the following advantages:

1. The present invention can automatically create a range of landmarks using web albums or photos uploaded by the public on a social networking site.

2. The present invention allows a user to see a range of landmarks on an electronic map by entering a keyword, rather than just a representative point or circle.

3. The present invention can effectively filter out scattered or low-coordinate coordinate positions in public upload photos.

4. The present invention can automatically detect possible famous landmarks using web albums or photos uploaded by the public on the social networking site.

5. The invention can automatically suggest possible famous landmarks and popular tourist attractions, and provide them for reference.

6. The invention is particularly suitable for automatically marking urban landmarks in a large area on an electronic map without manual intervention.

7. Using the text label information attached to the photo, the invention can be applied to the exploration of events, such as the exhibition of the Flower Expo.

The present invention is a method for discovering landmarks and their scopes from social network photos, and taking a photo database built on a social networking site as an example. The photo database contains an uploader account number (or a photographer name). ), shooting date, GPS coordinates (also known as GPS tags), and photo annotations and explanatory texts (also known as text labels).

Referring to the flowchart of the method for marking the landmark area range shown in FIG. 1 , when the user wants to find the area range of the landmark on the map, the present invention proposes a method for discovering landmarks and their scopes by using social network photos. Its execution steps are:

a. The landmark label input by the user and the text label of all the photos in the photo database are processed by the text label comparison 100, and the matching photo is processed by the GPS label projection 200 onto the map, thereby forming a description of the landmark location. Black and white image with a size of 600*800.

b. The black and white image of step a is subjected to Gaussian kernel space low-pass filtering 300 operation processing in units of 9*9, and converted into a gray-scale image describing the position of the landmark.

c. The grayscale image of step b is processed by image binarization 400 by a threshold value to become a black and white image describing the range of the landmark area, and the image represents the range of the area covered by the landmark map.

In step a, the GPS tag of the photo is projected 200 onto the map, and the formula used for coordinate position conversion is:

Among them, Lng is the longitude coordinate, Lat is the latitude coordinate, PIC_W is the width of the map and the black and white image describing the landmark position, PIC_H is the length, and (x, y) is the pixel coordinate. Take the regional coordinates in the map of Annex I as an example:

PIC _ W =600, PIC _ H =800;

Lat _min =24.94, Lat _max =25.14;

Lng _min = 121.46, Lng _max = 121.61.

See Attachment 1 for a black and white image depicting the location of the landmark and nesting the projections of the photos in all maps in the photo library (indicated by gray dots). It can be clearly seen that there are some unreasonable scattered points on the left side, and the areas marked in the middle are not continuous enough.

In step b, the Gaussian core used in spatial low-pass filtering 300 waves is:

Where s=5, Δi is x-x0, Δj is y-y0, and (x0, y0) is the center point of the Gaussian core.

In step c, the threshold used by the image binarization 400 is MAX(Ig)/8, where MAX(Ig) is the maximum pixel value in the grayscale image Ig after low pass filtering in the Gaussian core space. After the processing of steps b and c, the scattered or low-coordinate coordinate positions in the public upload photos can be effectively filtered out. These coordinate positions may be because some photos are taken from a distant building; or the text labels of the photos may contain Partial landmark keyword string caused by.

See Appendix 2 for a way to discover landmarks and their scope from social site photos, and use the photos collected on the social site to map the results of the landmark area on the map. It has been found that the present invention can effectively filter out the scattered points scattered in the original Annex I, and the marked areas become relatively continuous.

Please refer to the flow chart for exploring the famous landmark method shown in Figure 2. When users want to know which famous landmarks or tourist attractions are on the map, the steps are as follows:

i. Filter the text labels of all the photos in the photo database through the text label representative determination program 500 to select a representative text label.

Ii. The representative text labels selected in step i are searched for the black and white image of the range of the landmark area by the representative label of the landmark area range marking program 600.

Iii. The black and white images of all the landmark areas in step ii are passed through the landmark distribution range characteristic determination program 700 one by one, and the text labels represented by the conformees are judged to be famous landmarks.

Taking the area in the map of Annex 1 as an example, the photos with the GPS tags in the database are selected first, and the text labels of the photos are passed through the representative tag determination program 500 as follows. Text label:

(1) This text label appears in the photo database more than once.

(2) This text label appears in the photo database above a number of different shooting dates.

(3) This text label is provided by a plurality of different callers in the photo database.

The representative text labels selected in step i are used to obtain the black and white images describing the range of the landmark regions one by one according to the method of marking the landmark region range described above, and then the black and white images of each landmark region range are subjected to the following landmark distribution ranges. The characteristic determination program 700 determines that the text label represented by the compliant person is a famous landmark:

(1) The number of black areas in the black and white image of the landmark area is lower than a threshold;

(2) If there are more than two black areas in the black and white image of the landmark area, the black and white image of the area of the landmark area is divided into a plurality of squares of the same size, and the square of each black area At least adjacent to or overlapping with the square in which the other black area is located.

Please refer to Annex 3 for the black and white image of the landmark area of the two representative text labels of "A" and "B". After passing the test of 700 characteristics of the distribution criteria, you can find that "A" has In one black area, there are more than 20 black areas in "B", so it is judged that "A" is a famous landmark.

The detailed description of the present invention is intended to be illustrative of a preferred embodiment of the invention, and is not intended to limit the scope of the invention. The patent scope of this case.

To sum up, this case is not only innovative in terms of technical thinking, but also has many of the above-mentioned functions that are not in the traditional methods of the past. It has fully complied with the statutory invention patent requirements of novelty and progressiveness, and applied for it according to law. Approved this invention patent application, in order to invent invention, to the sense of virtue.

100. . . Text label comparison processing

200. . . GPS tag projection processing

300. . . Spatial low pass filtering

400. . . Image binarization

500. . . Text label representative decision procedure

600. . . Representative labeling its landmark area range marking procedure

700. . . Landmark distribution range characteristic determination procedure

Please refer to the detailed description of the present invention and the accompanying drawings, and the technical contents of the present invention and its effects can be further understood; the related drawings are:

1 is a flow chart of a method for marking a landmark area range according to the present invention;

Figure 2 is a flow chart of the method for discovering famous landmarks in the present invention.

[A brief description of the attachment]

Annex I is a black and white image of the landmark location in the map depicting the landmark area range method;

Annex II is a black and white image of the range of landmarks in the map depicting the method of marking the area of the landmark;

Annex III is a schematic diagram of the characteristics of the landmark distribution range for the discovery of famous landmark methods.

100. . . Text label comparison processing

200. . . GPS tag projection processing

300. . . Spatial low pass filtering

400. . . Image binarization

Claims (10)

A method for discovering a landmark and a range thereof from a photograph of a social network site, wherein the method for marking the range of the landmark region includes: first determining a landmark keyword, which is an instruction that the user has to input; secondly, confirming the photo database, which is a photo data. The file storage database; the text label comparison processing is performed by comparing the landmark keyword with the text label of all the photos in the photo database; and then the GPS label projection processing is to compare the text label. The conforming surface data is projected onto the map data via the GPS tag, and is formed into a basic black and white image describing the landmark position; and the spatial low-pass filter is used to calculate the basic black and white image as the image filtering slope. Processing, and then forming a grayscale image describing the location of the landmark; finally, through image binarization, the grayscale image is processed by image binarization into a black and white image depicting the range of the landmark region. A method for excavating a landmark and a range thereof from a social network photo, as described in claim 1, wherein the photo database includes a GPS tag and a text tag of the photo. A method for excavating a landmark and a range thereof from a social network photo, as described in claim 1, wherein the text label is a photo file name, a text in a photo header, an album title, a photo annotation or a description. Text. A method for excavating a landmark and a range thereof from a social network photo, as described in claim 1, wherein the spatial low pass filter is also a Gaussian core space low pass filter. A method for discovering famous landmarks includes: first confirming a text label, which is a text label of all photos in the photo database; and then determining a program through a representative text label, the text label is determined by the representative text label The program screens out the representative text labels; and then through the representative label, the landmark area range marking procedure is to find the black and white image of the landmark area range one by one for the representative text labels; and the landmark distribution range characteristic determination program is A determination procedure for examining the characteristics of the landmark distribution range of the representative character label; and the last famous landmark is the representative text label conforming to the landmark distribution range characteristic determination program. A method for excavating a landmark and a range thereof from a social network photo, as described in claim 5, wherein the photo database includes a GPS tag and a text tag of the photo. A method for excavating a landmark and a range thereof from a photograph of a social network website according to claim 5, wherein the representative process of the text label includes a combination of one or more conditions, and the conforming person determines the representative text. label. A method for excavating a landmark and a range thereof from a social network photo, as described in claim 6, wherein the condition is that the text label appears in the photo database for more than a plurality of times, and the text label is in the photo database. Any one of the plurality of different photographing dates appearing in the middle or the number of the text labels in the photo database is provided by the plurality of different labelers: A method for excavating a landmark and a range thereof from a photograph of a social network site according to claim 5, wherein the landmark distribution range characteristic determining program includes a combination of one or more conditions, and the conforming person determines that the representative text label is Famous landmarks. A method for excavating a landmark and a range thereof from a photograph of a social network site according to claim 8 of the patent application, wherein the condition is that the number of landmark areas in the black and white image of the landmark area is lower than a threshold or the range of the landmark area When there are more than two landmark areas in the black and white image, the black and white image of the landmark area is divided into a plurality of squares, and each of the landmark areas is located at least adjacent to the square of the other landmark area. Either or overlap is the landmark distribution range characteristic determination procedure.