View - based Object Representation using Graphs of Multi - scale Feature Descriptors

نویسندگان

  • Clas Jönsson
  • Tony Lindeberg
چکیده

Decomposition of image objects into a set of predefined structural primitives, or features, is an intuitive approach with image analysis for object representation. In this work we investigate how multi-scale shape features may be extracted and placed in relation to each other, to form an abstracted graph-description of image objects. The resulting graph representation will be a Directed Acyclic Graph (DAG), with the extracted features as nodes and the edges indicating overlap between the image features. To increase the amount of information held by the graph, the nodes and arcs are labeled with feature and relational characteristics. We apply this representation to a simplified recognition problem in the domain of hand gesture recognition. The objective is to match test graphs computed from images of hands in various poses to model graphs computed from images of hands in known poses. Vybaserad objektrepresentation med grafer av särdrag på multipla skalor Sammanfattning Uppdelning av bildobjekt i fördefinierade strukturella primitiver är ett intuitivt tillvägagångssätt vid objektrepresentation för datorseende. I detta arbete undersöker vi möjligheten att använda geometriska särdrag på multipla skalor, tillsammans med relationer mellan dessa, för att erhålla en grafrepresentation av bildobjekt. Den resulterande representationen blir en riktad acyklisk graf, med de extraherade särdragen som noder, och bågar införda där överlapp mellan dessa detekterats. För att öka informationsinnehållet i grafen associeras noderna och bågarna med attribut för särdragen respective relationerna de representerar. Vi tillämpar representationen i ett förenklat objektigenkänningssystem för handgestigenkänning. Målet är att matcha grafer genererade från testbilder på händer i olika poser mot grafer genererade från bilder av kända poser.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Affine Invariant Interesting Descriptors

Component trees or level set graphs have proved to be efficient tools for the extraction of stable features in many image analysis and understanding applications. In this paper, we explore the relation between level sets as feature descriptors and curvature scale-space. While most classical image feature detectors are based on linear scale-space, affine-invariant descriptors can be related to t...

متن کامل

Multi-layer Local Graph Words for Object Recognition

In this paper, we propose a new multi-layer structural approach for the task of object based image retrieval. In our work we tackle the problem of structural organization of local features. The structural features we propose are nested multi-layered local graphs built upon sets of SURF feature points with Delaunay triangulation. A Bag-of-Visual-Words (BoVW) framework is applied on these graphs,...

متن کامل

Many-to-Many Feature Matching Using Spherical Coding of Directed Graphs

In recent work, we presented a framework for many-to-many matching of multi-scale feature hierarchies, in which features and their relations were captured in a vertex-labeled, edge-weighted directed graph. The algorithm was based on a metric-tree representation of labeled graphs and their metric embedding into normed vector spaces, using the embedding algorithm of Matous̆ek [13]. However, the me...

متن کامل

Objects Recognition Using the Histogram Based on Descriptors of SIFT and SURF

This article reviews a method of histograms based on descriptors of SIFT and SURF and their application in Recognition of three-dimensional object recognition from various view. This method will be provided for Recognition of objects in a set of three-dimensional objects and images of ETH80 databases to assess the effectiveness of methods used in determining the threedimensional objects. Finall...

متن کامل

Using a Novel Concept of Potential Pixel Energy for Object Tracking

Abstract   In this paper, we propose a new method for kernel based object tracking which tracks the complete non rigid object. Definition the union image blob and mapping it to a new representation which we named as potential pixels matrix are the main part of tracking algorithm. The union image blob is constructed by expanding the previous object region based on the histogram feature. The pote...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2009