An Improved Shock Graph Approach for Shape Recognition and Retrieval

Abstract

Object recognition and shape matching are important issues in the field of image processing. Extraction and application of skeleton of a shape is widely used in these fields. In this paper shape matching and retrieval is performed using one of the skeleton-based methods called "shock graphs". By modifying and optimizing this method, results have been improved significantly, especially in the presence of occlusion and missing parts.In this extension, branch points are added as key points to the shock graph and its grammar and consequently a new grammar is developed. Our experimental results due to the modifications and extensions are presented with different examples and tests, especially in the presence of occlusion and missing parts.

Fast and Simple 2D Shape Retrieval Using Discrete Shock Graph

“…The second and fourth order shocks are the generic cases of shock orders which are involved in occurring instability in shape retrieval [2], [3]. The second order shocks are the sources of flow while the fourth order shocks are termination points of flow [1], which represent branch and end points, respectively.…”

“…According to them, instability occurs due to the junction points (or branch points) of the skeleton, and therefore, matching is only done with respect to end points. In contrast, Zaboli et al [3] and Goh [4] reduced time complexity using branch points only. However, considering only branch or only end points is not always accurate, because some images, like those of a circle, may not have any end points.…”

A novel graph-based matching method to merge the extracted maps from mobile robots

“…Some of these methods are: the ICP 1 algorithm and its subtype such as MBICP 2 , PLICP 3 and also the methods like IDC 4 . These algorithms usually estimate the difference of two laser scanner shots and from the obtained difference, the displacement ratio and the changes in direction of robot (if any) is calculated.…”

“…The method, proposed in this paper, is the method based on image processing that has higher speed than of those proposed by Karpin and his coworkers and also taking the rules of rescue robot into consideration. In papers [3] and [4] methods for obtaining the skeleton of the picture and also the graph from this skeleton has been introduced such that they are also used in the proposed algorithm here. The organization of this paper is as follows: In section 2 the description of obtaining the map by the robot is discussed and section 3 explains how the common parts of paths of various robots are calculated.…”

Inferência de gramáticas estocásticas para reconhecimento de padrões de imagens utilizando quadtrees

“…O sistema criadoé formado por duas partes principais: análise e reconhecimento. A análise consiste da seleção de primitivas (fragmentos de superfícies) Um shock graphé obtido a partir dos shock groups e são grafos direcionados e acíclicos[61].Shock graphs, um método baseado em esqueletos, foi utilizado em[62] para reconhecimento de objetos. O método utiliza pontos de ramificações (que têm 3 ou mais pontos vizinhos na vizinhança de 8 pixels) em shock graphs e shock grammars para melhorar a taxa de reconhecimento e a eficiência no reconhecimento de objetos.…”