-
Notifications
You must be signed in to change notification settings - Fork 3
/
kmean_omp.c
252 lines (223 loc) · 5.16 KB
/
kmean_omp.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
#include <omp.h>
#include <stdio.h>
#include <math.h>
#include <stdlib.h>
#include <time.h>
typedef struct
{
double _r;
double _g;
double _b;
double _m;
double _n;
} Point;
//Reads the image dimensions a x b pixels
void readImageSize(FILE *ifp,int* a,int* b)
{
fscanf(ifp,"%d\n",a);
printf("%d\n",*a);
fscanf(ifp,"%d\n",b);
printf("%d\n",*b);
}
//reads the ifp file and stores in structure
void readPoints(FILE* ifp,Point *points, int num_points)
{
int i;
for(i=0;i<num_points;i++)
{
fscanf(ifp,"%lf,%lf,%lf,%lf,%lf", &points[i]._r, &points[i]._g, &points[i]._b, &points[i]._m, &points[i]._n);
}
}
//Initialize random points as assumed means
void initialize(Point* mean,int K, int num_points, Point* points)
{
int i, a, p=2;
srand(time(NULL));
for(i=0;i<K;i++)
{
a = num_points/p;
mean[i]._r = points[a]._r;
mean[i]._g = points[a]._g;
mean[i]._b = points[a]._b;
mean[i]._m = points[a]._m;
mean[i]._n = points[a]._n;
p++;
}
}
//All points having no clusters
int IntClusterMem(int *cluster, int num_points)
{
int i;
for(i=0;i<num_points;i++)
{
cluster[i]=-1;
}
}
//Distance
double calculateDistance(Point point1,Point point2)
{
return sqrt((pow((point1._r-point2._r),2)+pow((point1._g-point2._g),2)+pow((point1._b-point2._b),2)));
}
//to calculate which cluster is the point belonging to.
int pointsCluster(Point point,Point* mean,int K)
{
int parent=0;
double dist=0;
double minDist=calculateDistance(point,mean[0]);
int i;
for(i=1;i<K;i++)
{
dist=calculateDistance(point,mean[i]);
if(minDist>=dist)
{
parent=i;
minDist=dist;
}
}
return parent;
}
//calculate new mean
void calcNewMean(Point* points,int* cluster,Point* mean,int K,int num_points)
{
Point* newMean=malloc(sizeof(Point)*K);
int* members=malloc(sizeof(int)*K);
int i;
for(i=0;i<K;i++)
{
members[i]=0;
newMean[i]._r=0;
newMean[i]._g=0;
newMean[i]._b=0;
newMean[i]._m=0;
newMean[i]._n=0;
}
for(i=0;i<num_points;i++)
{
members[cluster[i]]++;
newMean[cluster[i]]._r+=points[i]._r;
newMean[cluster[i]]._g+=points[i]._g;
newMean[cluster[i]]._b+=points[i]._b;
}
for(i=0;i<K;i++)
{
if(members[i]!=0.0)
{
newMean[i]._r/=members[i];
newMean[i]._g/=members[i];
newMean[i]._b/=members[i];
}
else
{
newMean[i]._r=0;
newMean[i]._g=0;
newMean[i]._b=0;
newMean[i]._m=0;
newMean[i]._n=0;
}
}
for(i=0;i<K;i++)
{
mean[i]._r=newMean[i]._r;
mean[i]._g=newMean[i]._g;
mean[i]._b=newMean[i]._b;
mean[i]._m=newMean[i]._m;
mean[i]._n=newMean[i]._n;
}
}
//check for convergence
// it checks that is each points cluster remaining the same
int chkConvrg(int *before_clusters,int *after_cluster,int num_points, float tol)
{
int i;
tol = num_points*tol;
for(i=0;i<num_points;i++)
if(abs(before_clusters[i]-after_cluster[i])>tol)
return -1;
return 0;
}
int main(int argc, char* argv[])
{
int K;
int num_points;
int i;
int job_done=0;
int x,y;
float tol;
struct timespec tic;
struct timespec toc;
double tspantemp = 0.0, tspan = 0.0;
Point* mean;
Point* points;
Point* get_points;
int * formed_clusters;
int * before_clusters;
int * after_cluster;
int thread_no;
K=atoi(argv[4]);
tol=0;
thread_no = atoi(argv[3]);
//Readinf file
FILE *ifp;
ifp=fopen(argv[1],"r");
readImageSize(ifp,&x,&y);
num_points = x*y;
points=(Point*)malloc(sizeof(Point)*num_points);
readPoints(ifp,points,num_points);
fclose(ifp);
clock_gettime(CLOCK_REALTIME, &tic);
double myclocktemp = (double)tic.tv_nsec;
before_clusters=(int*)malloc(sizeof(int)*num_points);
after_cluster=(int*)malloc(sizeof(int)*num_points);
mean=malloc(sizeof(Point)*K);
//initializing to default values
initialize(mean,K,num_points,points);
IntClusterMem(before_clusters,num_points);
IntClusterMem(after_cluster,num_points);
while(1)
{
omp_set_num_threads(thread_no);
#pragma omp parallel default(shared)
#pragma omp for
for(i=0;i<num_points;i++)
{
after_cluster[i]=pointsCluster(points[i],mean,K);
}
calcNewMean(points,after_cluster,mean,K,num_points);
printf("New Centroids are calculated!\n");
if(chkConvrg(after_cluster,before_clusters,num_points,tol)==0)
{
printf("K-mean algorithm Converged!\n");
job_done=1;
}
else
{
printf("Not converged!\n");
for(i=0;i<num_points;i++)
before_clusters[i]=after_cluster[i];
}
if(job_done==1)
break;
}
clock_gettime(CLOCK_REALTIME, &toc);
myclocktemp = (double)toc.tv_nsec;
if((toc.tv_nsec - tic.tv_nsec) > 0)
tspantemp = (double)(toc.tv_nsec - tic.tv_nsec)/1000000000.0;
else
tspantemp = (double)(tic.tv_nsec-toc.tv_nsec)/1000000000.0;
printf("Total time elapsed in forming clusters : %f sec\n", tspantemp);
//Outputting to the ofp file
FILE* ofp=fopen(argv[2],"w");
//fprintf(ofp,"%d\n",K);
fprintf(ofp,"%d\n",x);
fprintf(ofp,"%d\n",y);
for(i=0;i<K;i++)
fprintf(ofp,"%d,%d,%d,%d,%d\n",(int)mean[i]._r,(int)mean[i]._g,(int)mean[i]._b,(int)mean[i]._m,(int)mean[i]._n);
for(i=0;i<num_points;i++)
fprintf(ofp,"%d,%d,%d,%d,%d,%d\n",(int)points[i]._r,(int)points[i]._g,(int)points[i]._b,(int)points[i]._m,(int)points[i]._n,after_cluster[i]+1);
fclose(ofp);
FILE* timef=fopen("TIME/OMP_Time","a");
fprintf(timef,"%f\n",tspantemp);
fclose(timef);
//End of all
return 0;
}