FR2897183A1 - METHOD FOR VERIFYING THE SAVING AREAS OF A MULTIMEDIA DOCUMENT, METHOD FOR CREATING AN ADVERTISING DOCUMENT, AND COMPUTER PROGRAM PRODUCT - Google Patents

Abstract

The invention relates to a method for checking the salient areas of a multimedia document. According to the invention, the method comprises the steps of: - determining (E1) areas of said document considered as key areas from data a priori extracted from said document, - application (E2) to said document of a model based on the establishment of saliency maps to obtain a saliency map of said document, - comparison (E3) between the areas of said document considered as key and said saliency map.

Description

The invention relates to a device and a method for verifying

  salient areas of a multimedia document.

  The invention applies more particularly to the evaluation of content and more particularly to documents such as advertising documents. Today, when we want to highlight a content, various techniques of content creation are known: simply, when writing a text document, we highlight, we write in bold, we frame .. More sophisticated techniques are used by the publicists who can, for video or audio, combine image processing techniques or sound processing techniques, more generally signal processing techniques.

  It is difficult to evaluate the effectiveness of these techniques. Advertisements are often attractive but the various effects implemented are not always all effective and could certainly be improved. Currently, oculometric techniques can help in determining salient areas. Such techniques, however, are manual and cumbersome to implement because they do not allow automation of the method on a plurality of documents, web pages, video documents.

  For this purpose, the invention proposes a method for verifying the saliency zones of a multimedia document. According to the invention, the method comprises the steps of: - determining the areas of said document considered as key areas from data a priori extracted from said document, -application to said document of a model based on the establishment of saliency maps in order to obtain a saliency map of said document, comparison between the areas of said document considered as key and said saliency map.

  According to a preferred embodiment, the method comprises a step of calculating a ratio between the salience contained in the key areas of said document and the total saliency of said document.

  According to a preferred embodiment, the method comprises a step of calculating the visual path indicating the course of the eye during the display of said document, said visual path being calculated from said saliency map and composed of the pixels of the image whose saliency is greater than a threshold.

  According to a preferred embodiment, when searching for said visual path, the pixel having the maximum saliency in said document is searched for, - an area of said document is inhibited around said maximum and - the next pixel having the maximum saliency is searched for, until all the pixels having a saliency higher than said threshold have been obtained.

  Advantageously, an analysis report is established comprising the saliency zones, said visual path, and a set of sub-images representative of the saliency zones of said document.

  According to a preferred embodiment, said report comprises the key areas of said document. The invention also relates to a method of creating an advertising document. According to the invention, the method alternatively comprises a content creation step followed by a method for verifying the saliency zones of said document according to the invention, said steps being repeated as long as the saliency zones do not correspond to the key zones. said document.

  The invention also relates to a method for automatically checking the salient areas of a website. According to the invention, the method comprises, for each page of the site, the steps of - determining the zones of said page considered as key zones from the data extracted a priori from said page, - applying to said page a model based on the establishment of saliency maps to obtain a saliency map of said page, - comparison between the areas of said page considered as key and said saliency map. The invention also relates to a computer program product comprising program code instructions for performing the steps of the methods according to any one of the embodiments of the invention when said program is executed on a computer. The invention will be better understood and illustrated by means of examples of advantageous embodiments and implementations, in no way limiting, with reference to the appended FIG. 1 representing the verification method according to a preferred embodiment of the invention. The invention is particularly advantageous in advertising documents where the components of the document are particularly important because of their economic impact. Indeed, commercials, whether video, audio or simply still images in a magazine, are particularly worked to place messages attractive to the consumer. However sometimes these messages are not necessarily the areas that will attract the most the human eye and in this case, the publicist will not have created a document with the desired effect. The present invention thus aims to provide means for analyzing the documents, whether advertising or not, to verify that these documents have the desired impact to their recipients.

  Thus, there are methods of calculating saliency maps for a document. These methods make it possible to determine the areas that will attract the eye. They are based on calculations taking into account different parameters of the image and on spatial and / or temporal analyzes of the image and / or the video. A saliency map is a topographic representation of the perception of an image. The visual relevance of each pixel of the image is determined. The map is standardized for example between 0 and 255. The higher the value, the more the pixel attracts the eye.

  Such methods are described, for example, in the European patent application filed by Thomson Licensing SA on June 22, 2005 under number 1544792.

  Figure 1 shows the steps of the method according to a preferred embodiment of the invention. There is a paper document D of the commercial document type. This document was created from an editing software, for example acrobat writer (registered trademark by the company adobe), word (trademark registered by the company microsoft), type HTML, which is available electronic version. One can thus extract from this document a priori information which reflects what the author of the document wished to put forward. For example, in a word document, we can judge the importance of the information according to the size of the characters used, the font used, the highlighting, the color used. The location on the sheet can also be used, generally the important information being placed in the center of the image. In a html document, tags are also a help in determining the points that the author wants to put forward. Thus, manually by observing the document D, or automatically by automatic extraction of the characteristics mentioned above, it is possible to establish (step E1) the information a priori representative of the key points of the document D according to the author. It is thus possible to obtain the geographical location of the areas of interest of the document.

  On the other hand there is a model for building the saliency map of a document. During a step E2, this model is applied to the document D. The model is for example a software receiving the document D as input. The model analyzes the characteristics of the document necessary for the extraction of salient maps in order to obtain the D. Such a method is for example described in the European patent application on behalf of Thomson Licensing SA, published on June 22, 2005 under the number 1544792. They are based on calculations taking into account different image parameters and based on spatial and / or temporal analyzes of the image and / or video. Text type information as well as face type information is also extracted because usually they are also representative of information considered important when the text is included in a video document. Taking into account the information of the face or text type during the establishment of saliency maps is notably described in the document attention mode) based progressive image transmission of Hu, Xie, Chen, Ma as well as in the document a visual attention mode ) for 20 adapting images on small displays by Chen and Al and published in the journal multimedia systems in 2003.

  During a step E3, from the saliency map, an analysis report is obtained containing the following information: A heat map indicating the zones of attraction, A visual path map indicating the visual path of the average observer. This visual path is obtained by searching the saliency map for the local maxima having a saliency value greater than a threshold, as a function of the number of maxima that one wishes to obtain or as a function of the global saliency of the image. The search for local maxima can for example be done by inhibiting a circular area (representing the fovea of radius equal to 0.5 visual degree) around the (n-1) th maximum and looking for the global maximum of this new map and so on for each maximum. For each maximum one thus obtains a zone and then one connects the centers of zones n-1, n, ... in order to obtain the visual path of an average observer. - A perceptual summary containing n sub-images of the image to be analyzed. These sub-images are areas fixed by the observer. Such a method for determining the sub-images is described in the European patent application filed on September 19, 2005 under number 05291938.8 in the name of Thomson Licensing. This method is also described in the publication of O. Le Meur, X. Castellan, P. The Callet and D. Barba, Efficient saliency-based repurposing method, ICIP 2006. The sub-images are composed of pixels around the local maxima. The number of sub-images is determined by setting a threshold and sub-images are defined only for maxima having a saliency value greater than this threshold.

  This information makes it possible to detect the zones of the image actually perceived by the human eye.

  The areas indicated by the author of the document or automatically extracted during step E1 and the zones indicated in the analysis report are then compared. We are thus able to compare what the author wanted to put forward in the document to the really catchy information for the human eye, by putting in front of the information a priori and the information of the analysis report.

  The analysis report can also directly provide this analysis by implementing a step of direct comparison between the saliency map and the prior data. Indeed, one can compare the geographic location of salient areas and areas considered important by the author.

  For example, the ratio between the salience contained in the areas delimited by the author of the document and the overall saliency can be calculated. This report can be an indicator of the relevance of the document's design. It is thus possible for the creator to evaluate in terms of the attraction for the eye the value of his document.

  The invention therefore advantageously proposes a method for evaluating documents of a commercial or advertising nature.

  The invention can also be seen as a remote method for the evaluation of advertising documents. We can actually consider that the model is available on a website and that the customer submits his document in electronic form to know the points / areas that will actually hang the human eye. The user of the service, depending on the result of the comparison can then modify his document and resubmit it iteratively so that the information he wishes to put forward is really the information on which the human eye is going to go. focus.

  The method may also have as application the evaluation of several documents submitted by customers following a call for tenders. It is thus possible to analyze automatically and objectively what is the document that will have the greatest impact on consumers.

  In another embodiment, the document D is not a paper document or a still image, of the advertising insert type but a video document such as the advertisements broadcast on the television channels. Such a document can also be analyzed according to the method implemented in the invention.

  Indeed, it is also possible to establish a salience map from time information. Such a method is described in the document of 0. Le Meur, P. Le Callet, D. Barba and D. Thoreau, "A spatio-temporal model of the selective human visual attention", Proc. ICIP-05 (IEEE International Conference on Image Processing), Genoa, Italy, September 2005.

  We can thus also obtain the saliency map of the video document. To obtain the information on the video data, it is possible to obtain them via the author of the video.

  In another embodiment, this method can also be used by integrating it directly into document creation aids. Indeed, we obtain documents for which the eye focuses on the areas actually considered as key in the document. It is thus possible alternately to create content, to apply the verification method according to the invention and to modify it as long as the salient zones and the key zones do not coincide.

  In yet another embodiment, it may also be envisaged that the document to be analyzed is a website and that each page of the site is analyzed regularly, automatically or manually. Thus, one can envisage an automatic procedure for launching the method according to the invention on all the web pages of the site and display of the report, comparative result between the data considered as key in the page by the author and the data of the page actually visualized by the eye if available, or blind if key data is not available.

Claims (9)

claims   1. A method of verifying the saliency zones of a multimedia document, characterized in that it comprises the steps of: determining (El) zones of said document considered as key zones from data extracted from said document, application ( E2) to said document of a model based on the establishment of saliency maps in order to obtain a saliency map of said document, - comparison (E3) between the areas of said document considered as key and said saliency map.   2. Method according to claim 1 characterized in that it comprises a step of calculating a ratio between the salience contained in the key areas of said document and the total saliency of said document.   3. Method according to claims 1 or 2 characterized in that it comprises a step of calculating the visual path indicating the course of the eye during the display of said document, said visual path being calculated from said saliency map and composed of the pixels of the image whose saliency is greater than a threshold.   4. Method according to claim 3 characterized in that during the search for said visual path, the pixel with the maximum saliency is sought in said document, an area of said document is inhibited around said maximum and the next pixel is searched with maximum saliency. , until all pixels having a saliency greater than said threshold have been obtained.   5. Method according to one of the preceding claims characterized in that one establishes an analysis report comprising the zones depreciation, said visual path, and a set of sub-images representative of the saliency areas of said document.   6. Method according to claim 5 characterized in that said report comprises the key areas of said document.   7. A method for creating an advertising document characterized in that it comprises alternately a content creation step followed by a method for verifying the saliency areas of said document according to one of claims 1 to 6, said steps being repeated until the salient areas match the key areas of the document.   8. A method of automatically verifying the salient areas of a website characterized in that it comprises for each page of the site the steps of - determining the areas of said page considered as key areas from data a priori extracted from said page, - application to said page of a template based on the establishment of saliency maps to obtain a saliency map of said page, - comparison between the areas of said page considered as key and said saliency map.   9. Computer program product characterized in that it comprises program code instructions for performing the steps of the methods according to any one of claims 1 to 8, when said program is executed on a computer.