Sei sulla pagina 1di 13

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

FACULTY OF PRODUCTION AND SERVICES

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC


COLLEGE OF SYSTEMS ENGINEERING

SUBJECT: STRUCTURE OF DATA II

2013 - AREQUIPA

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

INDEX
SLEEVE NOTES ............................................................................................................ 1 INTRODUCTION ........................................................................................................... 2 INDEXOFCONTENTS .................................................................................................. 3

1. Development ............................................................................................. 4
1.1. Why use data structuresR* tree and B+ tree? ......................................................... 4 1.2. Code ....................................................................................................................... 5 1.2.1. Main .............................................................................................................. 5 1.2.2. Node Arbol R_Asterist.h ............................................................................... 6 1.2.3. R-Asterist.cpp .............................................................................................. 8 1.2.4. R-Asterist.h ................................................................................................. 9 Conclusions .....................................................................Error! Marcador no definido.12 References ...................................................................................................................... 13

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

INTRODUCTION

In this project " EFFICIENCY IN THE TWO-DIMENSIONAL GEOMETRIC SEARCH " here development a new way of space access; using afusion of two data structures; Tree B+ and Tree R* ; in the case of Tree B+ the global access is faster and the height half of tree is smallest and the case of TreeR*, is a data structure that use space methods ,in other words, to indexing multidimensional information. The project was worked with sub images of a main image, and we'll look for procedures that let us to know if an image point is inside or outside a selected image, to carry out a searching effective in a map or an image according with x and y axes .

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

2. DEVELOPMENT
Our project has been developed in Visual Studio 2012, and the image handle has been worked with the library: SDL. We have worked with sub images of a main image, and searching procedure that let us to know if an image point is inside or outside a chosen image, and can identify: TOTAL INTERSECTION ANDPARTIAL.

1.1.

Why use data structures R * Tree and B+ Tree?

We use the B+ tree,because it is a kind of tree data structure, which represents a collection of tidydata, which let us an efficient insertion and deletion of elements. Another benefit that gives us the B+ tree is that global access is faster and the height half of the tree is smaller. Also its leaves can be linked, achieving of this way, a trajectory sequential to cover the keys in the tree. Another reason why we used the tree B + is when we want delete a key is relatively simple, because when we want to delete a key that is relatively simple ; because the keys eliminate is found in the pages sheets and the ability to leave the values without key in the part of indexes as separators simplifies the process . Also to carry out our project is used the tree R*, because it is a structure of data of type tree ,that used structure to methods of spatial access , in other words , for indexing multidimensional data , for example , the axis ( x , y) a geographical location . The R* tree divides the space of way hierarchically in collection, possibly overlapping. Next the development of our project EFFICIENCY IN THE TWO-

DIMENSIONAL GEOMETRIC SEARCH ".

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

1.2.

Code

1.2.1. Main
#include "SDL.h" #include <string> #include <iostream> #include "SDL_image.h" #include <SDL_image.h> #include "Clase_Imagen.h" #include "NodeArbolR_Asterist.h" #include "R_Asterist.h" #define FONDO_IMAG "Mundo2.bmp" #define ORDEN 5 using namespace std; SDL_Surface *pantalla; void esperar(void); SDL_Surface* recortarImag(SDL_Surface *tobecropped, SDL_Rect *r); int main(int args, char* argv[]){ pantalla = new SDL_Surface(); SDL_Init(SDL_INIT_EVERYTHING); Clase_Imagen<int> *obj = new Clase_Imagen<int>(); obj->cargarImagen(FONDO_IMAG); obj->mostrarImag(pantalla,obj->getImagen()); R_Asterist< Clase_Imagen<int>> *raiz = new R_Asterist< Clase_Imagen<int>>(ORDEN); int i = -1; for(; i != 0 ;){ cout<<"DESEA INSERTAR UNA NUEVA IMAGEN?"; cin>>i; swicth(i) case 0: SDL_QUIT(); case 1: Clase_Imagen<int> *obj2 = new Clase_Imagen<int>(); obj2->crearRectangulo(); SDL_Surface* newImag = obj->recortarImag(obj->getImagen(),&obj2->rect); obj2->setImagen(newImag); raiz->insertElement(*obj2); default : break; } while(true){ raiz->mostrar(); } esperar(); return 0; } void esperar (void) { SDL_Event evento; while ( SDL_WaitEvent (& evento)) { if (evento.type == SDL_QUIT || evento.key.keysym.sym == SDLK_ESCAPE) return; } }

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

1.2.2. NodeArbolR_Asterist.h
#pragma once template<class X> class NodeArbolR_Asterist { private: int orden; int cantElements; bool isFull; bool esHoja; X* elements; NodeArbolR_Asterist<X> *hijos; int ubicarPosicionInsercion(X); public: NodeArbolR_Asterist(void); NodeArbolR_Asterist(int); bool buscarEnNodo(X); bool insertElement(X); void mostrar(void); int getPosicionDeHijo(X); NodeArbolR_Asterist<X>* getHijos(void); X* getElements(void); int getCantElements(void); void setCantElements(int); bool getIsFull(void); bool vaciarNodo(void); ~NodeArbolR_Asterist(void); };

template<class X> NodeArbolR_Asterist<X>::NodeArbolR_Asterist(void){ isFull = false; esHoja = true; orden = 0; cantElements = 0; elements = NULL; hijos = NULL; } template<class X> NodeArbolR_Asterist<X>::NodeArbolR_Asterist(int orden){ isFull = false; esHoja = true; this->orden = orden; cantElements = 0; elements = new X[this->orden -1]; hijos = new NodeArbolR_Asterist<X>[this->orden]; } template<class X> NodeArbolR_Asterist<X>* NodeArbolR_Asterist<X>::getHijos(void){ return hijos; } template<class X> X* NodeArbolR_Asterist<X>::getElements(void){ return elements; } template<class X> void NodeArbolR_Asterist<X>::setCantElements(int cant){ cantElements = cant;

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

} template<class X> int NodeArbolR_Asterist<X>::getCantElements(void){ return cantElements; } template<class X> bool NodeArbolR_Asterist<X>::getIsFull(void){ return isFull; } template<class X> bool NodeArbolR_Asterist<X>::vaciarNodo(void){ isFull = false; int tempcant = cantElements; NodoBmas<X> *temp = NULL; for(int i = 0; i < cantElements; i++){ elements[i] = NULL; tempcant--; } cantElements = tempcant; //cout<<"\nElementos: " << cantElements<<"\n"; return true; } template<class X> int NodeArbolR_Asterist<X>::ubicarPosicionInsercion(X elem){ //Busqueda Linear "MEJORAR BUSQUEDA DE UBICACION" for(int i = 0; i < cantElements ;i++){ if(this->elements[i] >= elem) return i; } return cantElements; } template<class X> int NodeArbolR_Asterist<X>::getPosicionDeHijo(X elem){ int tempPos = ubicarPosicionInsercion(elem); return tempPos; } template<class X> bool NodeArbolR_Asterist<X>::insertElement(X elem){ X tempElem; X temp2 = elem; NodeArbolR_Asterist<X> *temphijo1 = NULL; NodeArbolR_Asterist<X> *temphijo2 = NULL; int tempCant = cantElements; bool retornar = false; if(!isFull){ if(cantElements == 0){ elements[0] = elem; retornar = true; }else{

for(int i = 0; i < tempCant; i++){

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

if(temp2 < this->getElements()[i]){ tempElem = elements[i]; //temphijo1 = &getHijos()[i]; elements[i] = temp2; //if(temphijo2) // getHijos()[i] = *temphijo2; temp2 = tempElem; //temphijo2 = temphijo1; } } elements[cantElements] = temp2; retornar = true; } if(retornar){ cantElements++; if(cantElements == orden-1) isFull = true; } } return retornar; } template<class X> bool NodeArbolR_Asterist<X>::buscarEnNodo(X elem){ for(int i = 0; i < cantElements ; i++){ if(this->elements[i] == elem) return true; else if(this->elements[i] > elem) return false; } return false; } template<class X> void NodeArbolR_Asterist<X>:: mostrar(void){} template<class X> NodeArbolR_Asterist<X>::~NodeArbolR_Asterist(void){}

1.2.3. R_Asterist.cpp
#include "R_Asterist.h" R_Asterist::R_Asterist(void) {} R_Asterist::~R_Asterist(void) {}

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

1.2.4. R_Asterist.h
#pragma once #include "NodeArbolR_Asterist.h" #include <queue> template<class X> class R_Asterist { private: int orden; int altura; NodeArbolR_Asterist<X> *raiz; NodeArbolR_Asterist<X>* dividirNodo(NodeArbolR_Asterist<X>*, NodeArbolR_Asterist<X>*); bool insertarEnPadre(NodeArbolR_Asterist<X>*, NodeArbolR_Asterist<X>*); bool dividirNodo(NodeArbolR_Asterist<X>*,X); NodeArbolR_Asterist<X>* OrdenAlfaHojas(NodeArbolR_AsteristX>*, X, int); NodeArbolR_Asterist<X>* OrdenAlfaNodos(NodeArbolR_Asterist<X>*, NodeArbolR_Asterist<X>*, int); NodeArbolR_Asterist<X>* OrdenBeta(NodeArbolR_Asterist<X>*, X, int); NodeArbolR_Asterist<X>* buscarNodoInsert(X); NodeArbolR_Asterist<X>* buscarNodoPadre(NodeArbolR_Asterist<X>*, X); public: bool insertElement(X); bool buscarElement(X); bool eliminarElement(X); void mostrar(void); R_Asterist(void); R_Asterist(int); ~R_Asterist(void); }; template<class X> R_Asterist<X>::R_Asterist(void){ orden = 0; altura = 0; raiz = NULL; } template<class X> R_Asterist<X>::R_Asterist(int n){ orden = n; altura = 0; raiz = new NodoBmas<X>(orden); } template<class X> NodeArbolR_Asterist<X>* R_Asterist<X>::buscarNodoInsert(X elem){ NodeArbolR_Asterist<X> *temp = raiz; int tempPos = temp->getPosicionDeHijo(elem); while(temp->getHijos()[tempPos].getHijos()){ temp = &temp->getHijos()[tempPos]; tempPos = temp->getPosicionDeHijo(elem); }return temp;}

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

template<class X> bool R_Asterist<X>::insertElement(X elem){ bool retornar = false; NodeArbolR_Asterist<X> *tempNodoInsert = this->buscarNodoInsert(elem); if(!tempNodoInsert->insertElement(elem)){ if(tempNodoInsert->getIsFull()){ dividirNodo(tempNodoInsert,elem); } } return retornar; } template<class X> NodeArbolR_Asterist<X>* R_Asterist<X>::OrdenAlfaHojas(NodeArbolR_Asterist<X>* node, X elem, int pos){ NodeArbolR_Asterist<X> *nodeInz = new NodoBmas<X>(orden); NodeArbolR_Asterist<X> *nodeDerc = new NodoBmas<X>(orden); int limite = orden % 2 +1; int posTemp = 0; if( pos <= limite){ nodeInz->insertElement(elem); }else{ nodeDerc->insertElement(elem); } int u = 0; for(; u <= limite; u++){ if( nodeInz->getCantElements() <= limite){ nodeInz->insertElement(node->getElements()[u]);}} u--; for(; u < node->getCantElements(); u++){ nodeDerc->insertElement(node->getElements()[u]);} node->vaciarNodo(); node->insertElement(nodeInz->getElements()[nodeInz->getCantElements()-1]); node->getHijos()[0] = *nodeInz; node->getHijos()[1] = *nodeDerc; return node; } template<class X> NodeArbolR_Asterist<X>* R_Asterist<X>::OrdenAlfaNodos(NodeArbolR_Asterist<X>* padre, NodeArbolR_Asterist<X>* node, int pos){ NodeArbolR_Asterist<X> *tempNode = new NodeArbolR_Asterist<X>(orden+1); NodeArbolR_Asterist<X> *newNode = new NodeArbolR_Asterist<X>(orden); /*tempNode->getHijos()[pos] = tempHijos[0];*/ tempNode->insertElement(node->getElements()[0]); NodeArbolR_Asterist<X> *tempHijos = node->getHijos(); *node = tempHijos[0]; tempNode->getHijos()[pos + 1] = tempHijos[1]; int posY = 0; for(int i = 0; i <= orden ;i++ ){ tempNode->insertElement(padre->getElements()[posY]); if( i != pos +1 ){ tempNode->getHijos()[i] = padre->getHijos()[posY]; posY++;}}

10

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

NodeArbolR_Asterist<X> *tempIzq = new NodoBmas<X>(orden); NodeArbolR_Asterist<X> *tempDerc= new NodoBmas<X>(orden); int limite = tempNode->getCantElements()/2; int p = limite +1; int x = 0; for(int i = 0; i <= tempNode->getCantElements(); i++){ if(i <= limite){ if( i != limite){ tempIzq->insertElement(tempNode->getElements()[i]); }else{ padre->vaciarNodo(); padre->insertElement(tempNode->getElements()[i]); } tempIzq->getHijos()[i] = tempNode->getHijos()[i]; }else{ if(i < tempNode->getCantElements()){ tempDerc->insertElement(tempNode->getElements()[i]);} tempDerc->getHijos()[x] = tempNode->getHijos()[i]; x++;}} padre->getHijos()[0] = *tempIzq; padre->getHijos()[1] = *tempDerc; int u = 0; return padre; } template<class X> NodeArbolR_Asterist<X>* R_Asterist<X>::OrdenBeta(NodeArbolR_Asterist<X>* node, X elem, int pos){ NodeArbolR_Asterist<X> *nodeDerch = new NodeArbolR_Asterist<X>(orden); for(int = 0; ){} return NULL; } template<class X> bool R_Asterist<X>::insertarEnPadre(NodeArbolR_Asterist<X>* padre, NodeArbolR_Asterist<X>* node){ NodeArbolR_Asterist<X>*tempHijos = node->getHijos(); if(padre->insertElement(node->getElements()[0])){ int pos = padre->getPosicionDeHijo(node->getElements()[0]); NodeArbolR_Asterist<X> *temp = new NodeArbolR_Asterist<X>(orden); *temp = padre->getHijos()[pos+1]; *node = tempHijos[0]; padre->getHijos()[pos+1] = tempHijos[1]; for(int i = pos+2; i < padre->getCantElements()+1;i++){ NodeArbolR_Asterist<X> *temp2 = new NodeArbolR_Asterist<X>(orden); *temp2 =padre->getHijos()[i]; padre->getHijos()[i] = *temp; temp = temp2; } return true;} return false;} template<class X> bool R_Asterist<X>::dividirNodo(NodeArbolR_Asterist<X>* node,X elem){ int pos = node->getPosicionDeHijo(elem); node = OrdenAlfaHojas(node,elem,pos); NodeArbolR_Asterist<X> *padre = buscarNodoPadre(node,node->getElements()[0]); //padre->mostrar();cout<< " PADRE "; if(node == padre){ return true; } NodeArbolR_Asterist<X> *tempNode = new NodeArbolR_Asterist<X>(orden); NodeArbolR_Asterist<X> *tempPadre = new NodeArbolR_Asterist<X>(orden);

11

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

CONCLUSIONS

We conclude thatthe use of theR *treeis more efficientto searcha locationby coordinatescompared to othertrees wherewe may losetoo much time.

It also offersoccupy a spaceefficient memorythis occurs whenthere is completeor partialintersection ofimageswhere a newsubdivisionoccurs andthe images

areoriginalbutineachthere isanyintersectionandthis helps us tonot storegarbageand occupyin memory. It also offersmemory savingsbecause it is notnecessary to loadthe entireimage tosearchbut onlychildren orsectorsof images thatare chosenforcritical operations.

12

SEARCHING EFFICIENCY IN TWO-DIMENSIONAL GEOMETRIC

REFERENCES

Antonin Guttman: R-Trees: A Dynamic Index Structure for Spatial Searching, Proc. ACM SIGMOD International Conference on Management of Data, ISBN 0-89791-128-8

Lars Arge, Mark de Berg, Herman J.Haverkort, Ke Yi: The Priority RTree: A Practically Efficient and Worst-Case Optimal R-Tree, Proc. ACM SIGMOD international conference on Management of data, ISBN 158113-859-8

http://es.wikipedia.org/wiki/%C3%81rbol-B%2B http://www.seanster.com/BplusTree/BplusTree.html http://www.slideshare.net/cesarpa/arbol-b http://www.ramos.utfsm.cl/doc/860/sc/ED-Cap55BTrees12010.pdf http://es.wikipedia.org/w/index.php?title=%C3%81rbol-R&action=edit

13

Potrebbero piacerti anche