dbscan.rar_DBSCAN_DBScan java

/*------------------------------------------- DBSCAN Algo implementation -----------------------------------------/ * *Currently supports only int entries. *Inputs are used from file "newfile1" and output is sent to "output" file. the input is in the form of columns separated by a constant number of white characters like in an excel sheet. * *Developes clusters based on density i.e all elements within a distance Eps = K_Distance are segregated into a single cluster. *Initially all points are considered to be belonging to the distinct clusters. Then elements within the Eps distance are merged into the same cluster, for every such action all elements given clusters numbers uptil now are tested to see if they are to be merged into this cluster only. * The output is read out by the MATLAB file and plotted out as distinct clusters. *Eps calculation can also be done by the similar code, EpsCalc.java */ import java.io.*; import java.util.*; import java.lang.Number.*; /** * * @author Saurabh */ public class Main { final int K_Factor= 20; final int K_Distance = 15; // K_Distance will be calculate in the another code final int MAXROW = 178; // Represents max row count int Clust_no = 1; // Represents cluster number int Border_Clust = 500; int DistVector[][] = new int[200][200]; Queue myqueue = new LinkedList(); // To store vertices to be processed next Integer X, Y; // Represent X and Y coordinates int i = 0; int Elements[][]= new int[MAXROW][5]; int Dist[]= new int[MAXROW]; /* File_Init function reads the data in the file into a 5 elements array. Column 1 and 2 represent the X and Y coordinate fo the element, 3 represents Distance to its nearest neighbour and 4 represents the Cluster number, 5 is left for future use*/ void File_Init() { try { FileReader reader = new FileReader("newfile1"); // File with all the input Scanner in = new Scanner(reader); // PrintWriter out = new PrintWriter (array[5]+"_dis.txt"); String line = in.nextLine(); while (in.hasNext()) { System.out.println("Going to read"); line = in.nextLine(); System.out.println(line); String [] Line_Token = line.split(" "); // Token delimiter is considered to be " " System.out.println(Line_Token[1]); X = Integer.parseInt(Line_Token[1]); // Column number to be used Y = Integer.parseInt(Line_Token[8]); // Column number to be used // System.out.println(X); // For testing purposes Elements[i][0]=X; Elements[i][1]=Y; i++; } }catch(Exception e) { System.out.println(e); } } /* Func_Dist function finds the minimum eauclidean distance between 2 points in 2 -d space*/ int Func_Dist(int X1, int Y1, int X2, int Y2) { int distance = (int)(Math.sqrt(((X2-X1)*(X2-X1)) + ((Y2-Y1)*(Y2-Y1)))); return distance; } /* Nearestk function finds the distance of the Kth closest neighbour for every elements and populates this in the 3rd column of the Elements array. *This code can be used for calculating the K_Distance factor value. */ /* int NearestK(int X1, int Y1, int K) { int Need =0; // Represents the need to process myqueue at the end of this for or not for(int k=0; k<200; k++) { Dist[k]= Func_Dist(X1, Y1, Elements[k][0], Elements[k][1]); } Arrays.sort(Dist); return Dist[K_Factor]; } */ /* *fincNum function helps in detemrining which all elements uptil now have been given the cluster number Cnumber. returns its index number */ int findNum(int Cnumber, int init) { int count = init; while((count >0)&&(Elements[count][3]!=Cnumber)) count--; return count; } /* Calc_K_dist function outputs the file and makes the clusters */ void Calc_K_dist() { try{ PrintWriter out = new PrintWriter ("c:/MATLAB7/work/outfile.txt"); for(int i=0; i<MAXROW; i++) Elements[i][3]=i; // Initialise cluster field of all elements to zero Elements[0][3]=Clust_no; for(int i=0; i<MAXROW; i++) // Makes a matrix DistVector and fills distance between all elements into this for(int j=0; j<MAXROW; j++) { DistVector[i][j] = Func_Dist(Elements[i][0], Elements[i][1], Elements[j][0], Elements[j][1]); } int tempState=0; int temp=tempState; for(int i=0; i<MAXROW; i++) { //Elements[i][2]= NearestK(Elements[i][0],Elements[i][1], Elements[i][3]); //Find and store K nearest distance for(int j =i; j>0; j--) if((DistVector[i][j] < K_Distance)&&(Elements[j][3]!=Elements[i][3])) { if(Elements[j][3]==0) Elements[j][3] = Elements[i][3]; // If element has not been given a cluster number uptil now, give it the same cluster number as i else { int tempvar=i, tempvar2, tempclust; tempclust = Elements[j][3]; // else give all the elements in the cluster of i the new cluster number while(tempvar >= 0) { tempvar2 = findNum(tempclust, tempvar); if(Elements[tempvar2][3] == tempclust) { //tempvar = (tempvar2)-1; Elements[tempvar2][3]= Elements[i][3]; } tempvar=(tempvar2)-1; } tempState = Elements[j][3]; temp = j; //Elements[j][3] = Elements[i][3]; } } // Give element J same cluster number as I else if((DistVector[i][j] == K_Distance)&&(Elements[j][3]==0)) // Assign J as I's border element Elements[j][3] = Border_Clust; else if(Elements[j][3