-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathkerneldefs.js
398 lines (356 loc) · 13.2 KB
/
kerneldefs.js
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
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
//gpu variable and functions
var gpu = new GPU();
function sqr(x) {
return x*x;
}
function dist(x1,y1,x2,y2) {
return Math.sqrt( sqr(x2-x1)+sqr(y2-y1) );
}
gpu.addFunction(sqr);
gpu.addFunction(dist);
//make rendering function
function makeRender() {
var img = gpu.createKernel(function(A) {
this.color(A[0][this.thread.y][this.thread.x],A[1][this.thread.y][this.thread.x],A[2][this.thread.y][this.thread.x]);
}).dimensions([800, 600]).graphical(true);
return img;
}
var toimg1 = makeRender();
/*START OF MAKE FILTERING FUNCTIONS*/
//default function to return image as array
//
function makeAnim(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var y = gpu.createKernel(function(img) {
return img[this.thread.z][this.thread.y][this.thread.x];
}, opt);
return y;
}
//make animation function, takes in a incremented position of x for each thread and return them
function makeAnimator(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A,x) {
return A[this.thread.z][this.thread.y][(this.thread.x + x)];
},opt);
return filt;
}
//3x3kernel sobel filter, perform 2d convolution twice, once for horizontal edge detection and another for vertical, before taking the square rooted squared sum.
/*
Convolution on input using 2 3x3 kernel each
x axis sobel kernel
[-1,0,1;
-2,0,2;
-1,0,1]
y axis sobel kernel
[1,2,1;
0,0,0;
-1,-2,-1]
lastly, to combine the vertical and horizontal convolution with:
squareroot((result1)^2 + (result2)^2)
*/
function makeSobelFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
//x axis sobel edge convolution
var c = A[this.thread.z][this.thread.y-1][this.thread.x-1]*-1 +
A[this.thread.z][this.thread.y][this.thread.x-1]*-2 +
A[this.thread.z][this.thread.y+1][this.thread.x-1]*-1 +
A[this.thread.z][this.thread.y-1][this.thread.x+1] +
A[this.thread.z][this.thread.y][this.thread.x+1]*2 +
A[this.thread.z][this.thread.y+1][this.thread.x+1];
//y axis sobel edge convolution
var d = A[this.thread.z][this.thread.y-1][this.thread.x-1]*-1 +
A[this.thread.z][this.thread.y-1][this.thread.x]*-2 +
A[this.thread.z][this.thread.y-1][this.thread.x+1]*-1 +
A[this.thread.z][this.thread.y+1][this.thread.x-1] +
A[this.thread.z][this.thread.y+1][this.thread.x]*2 +
A[this.thread.z][this.thread.y+1][this.thread.x+1];
//combining of the result
return Math.sqrt(Math.pow(c,2)+Math.pow(d,2));
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*gaussian blur 5x5 kernel to filter image
takes into account of more neighbouring thread to give higher level of feature than a 3x3 kernel
but has more multiplication operation per thread, 15 multiplication cost per pixel
[1,4,6,4,1;
4,16,24,16,4;
6,24,36,24,6;
4,16,24,16,4;
1,4,6,4,1] *(1/256)
*/
function makeBlurFilter5x5(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 1 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >1 && this.thread.z <3) {
//1st row
var c = A[this.thread.z][this.thread.y-2][this.thread.x-2]*1 +
A[this.thread.z][this.thread.y-2][this.thread.x-1]*4+
A[this.thread.z][this.thread.y-2][this.thread.x]*6+
A[this.thread.z][this.thread.y-2][this.thread.x+1]*4+
A[this.thread.z][this.thread.y-2][this.thread.x+2]*1+
//2nd row
A[this.thread.z][this.thread.y-1][this.thread.x-2]*4 +
A[this.thread.z][this.thread.y-1][this.thread.x-1]*16+
A[this.thread.z][this.thread.y-1][this.thread.x]*24+
A[this.thread.z][this.thread.y-1][this.thread.x+1]*16+
A[this.thread.z][this.thread.y-1][this.thread.x+2]*4+
//3rd row
A[this.thread.z][this.thread.y][this.thread.x-2]*6 +
A[this.thread.z][this.thread.y][this.thread.x-1]*24+
A[this.thread.z][this.thread.y][this.thread.x]*36+
A[this.thread.z][this.thread.y][this.thread.x+1]*24+
A[this.thread.z][this.thread.y][this.thread.x+2]*6+
//4th row
A[this.thread.z][this.thread.y+1][this.thread.x-2]*4 +
A[this.thread.z][this.thread.y+1][this.thread.x-1]*16+
A[this.thread.z][this.thread.y+1][this.thread.x]*24+
A[this.thread.z][this.thread.y+1][this.thread.x+1]*16+
A[this.thread.z][this.thread.y+1][this.thread.x+2]*4+
//5th row
A[this.thread.z][this.thread.y+2][this.thread.x-2]*1 +
A[this.thread.z][this.thread.y+2][this.thread.x-1]*4+
A[this.thread.z][this.thread.y+2][this.thread.x]*6+
A[this.thread.z][this.thread.y+2][this.thread.x+1]*4+
A[this.thread.z][this.thread.y+2][this.thread.x+2]*1;
return c/256;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
//optimized 5x5 gaussian blur, not working as intended
function makeBlurFilter5x5Optimized(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A,B) {
if (this.thread.y > 1 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >1 && this.thread.z <3) {
B[this.thread.z][this.thread.y][this.thread.x]= A[this.thread.z][this.thread.y+2][this.thread.x]*1+
A[this.thread.z][this.thread.y+1][this.thread.x]*4+
A[this.thread.z][this.thread.y][this.thread.x]*6+
A[this.thread.z][this.thread.y-1][this.thread.x]*4+
A[this.thread.z][this.thread.y-2][this.thread.x]*1;
return (B[this.thread.z][this.thread.y][this.thread.x-2]*1
+B[this.thread.z][this.thread.y][this.thread.x-1]*4
+B[this.thread.z][this.thread.y][this.thread.x]*6
+B[this.thread.z][this.thread.y][this.thread.x+1]*4
+B[this.thread.z][this.thread.y][this.thread.x+2]*1)/256;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*gaussian blur using 3x3 kernel convolution, total of 9 multiplication cost
convolution on 3x3 kernel on input array to give a blurring effect
[1/16,1/8,1/16;
1/8,1/4,1/8;
1/16,1/8,1/16]
*/
function makeBlurFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
var c = A[this.thread.z][this.thread.y-1][this.thread.x-1]*0.0625 +
A[this.thread.z][this.thread.y][this.thread.x-1]*0.125 +
A[this.thread.z][this.thread.y+1][this.thread.x-1]*0.0625 +
A[this.thread.z][this.thread.y-1][this.thread.x+1]*0.0625 +
A[this.thread.z][this.thread.y][this.thread.x+1]*0.125 +
A[this.thread.z][this.thread.y+1][this.thread.x+1]*0.0625+
A[this.thread.z][this.thread.y-1][this.thread.x]*0.125+
A[this.thread.z][this.thread.y+1][this.thread.x]*0.125+
A[this.thread.z][this.thread.y][this.thread.x]*0.25;
return c;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*attempt to optimize the 3x3 gaussian blur kernel, 6 multiplication cost in total
optimized version of gaussian blur, using 2 kernel filter function,
1 dimension convolution performed twice using a 3x1 kernel and a 1x3 kernel, 6 multiplication cost
1D vertical convolution with 3x1 kernel:
[0.25;
0.5;
0.25]
*/
//first part, vertical convolution, 3 multiplication cost
function makeBlurFilterOptimized(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
return A[this.thread.z][this.thread.y+1][this.thread.x]*0.25+
A[this.thread.z][this.thread.y][this.thread.x]*0.5+
A[this.thread.z][this.thread.y-1][this.thread.x]*0.25;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*second part, perform horizontal convolution with 1D on output of first part
1D horizontal convolution with 3x1 kernel:
[0.25 0.5 0.25;]
3 multiplication cost
*/
function makeBlurFilterOptimized_2(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(B) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
return B[this.thread.z][this.thread.y][this.thread.x-1]*0.25+B[this.thread.z][this.thread.y][this.thread.x]*0.5+B[this.thread.z][this.thread.y][this.thread.x+1]*0.25;
} else {
return B[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*Laplacian Sharpening filter using 3x3 kernel convolution:
[0 -1 0;
-1 5 -1;
0 -1 0]
gives a sharpening effect to image.
*/
function makeSharpFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
return A[this.thread.z][this.thread.y][this.thread.x-1]*-1 +
A[this.thread.z][this.thread.y][this.thread.x+1]*-1 +
A[this.thread.z][this.thread.y-1][this.thread.x]*-1 +
A[this.thread.z][this.thread.y+1][this.thread.x]*-1 +
A[this.thread.z][this.thread.y][this.thread.x]*5;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*Emboss filter using 3x3 kernel convolution:
[-2 -1 0;
-1 1 1;
0 1 2]
gives emboss effect
*/
function makeEmbossFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
var c = A[this.thread.z][this.thread.y-1][this.thread.x-1]*-2 +
A[this.thread.z][this.thread.y-1][this.thread.x+1]*0 +
A[this.thread.z][this.thread.y+1][this.thread.x-1]*0 +
A[this.thread.z][this.thread.y+1][this.thread.x+1]*2 +
A[this.thread.z][this.thread.y][this.thread.x-1]*-1 +
A[this.thread.z][this.thread.y][this.thread.x+1]*1 +
A[this.thread.z][this.thread.y-1][this.thread.x]*1 +
A[this.thread.z][this.thread.y+1][this.thread.x]*-1 +
A[this.thread.z][this.thread.y][this.thread.x]*1;
return c;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
/*filter that make image 'blink' at random interval
pass in Math.Random() added to each pixel
*/
function makeTVFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A,value) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
return A[this.thread.z][this.thread.y][this.thread.x]+value;
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}
//grayscale filter implemented by 0.2989*R+0.587*G+0.114*B
function makeGrayScalelFilter(mode) {
var opt = {
dimensions: [800, 600, 4],
debug: true,
graphical: false,
outputToTexture: true,
mode: mode
};
var filt = gpu.createKernel(function(A) {
if (this.thread.y > 0 && this.thread.y < 600-2 && this.thread.x < 800-2 && this.thread.x >0 && this.thread.z <3) {
return 0.2989 *A[0][this.thread.y][this.thread.x]+0.5870 *A[1][this.thread.y][this.thread.x]+0.1140 *A[2][this.thread.y][this.thread.x];
} else {
return A[this.thread.z][this.thread.y][this.thread.x];
}
},opt);
return filt;
}