iOS中的图像处理（二）——卷积运算

关于图像处理中的卷积运算，这里有两份简明扼要的介绍：文一，文二。

其中，可能的一种卷积运算代码如下：

- (UIImage*)applyConvolution:(NSArray*)kernel
{
	CGImageRef inImage = self.CGImage;
	CFDataRef m_DataRef = CGDataProviderCopyData(CGImageGetDataProvider(inImage));
	CFDataRef m_OutDataRef = CGDataProviderCopyData(CGImageGetDataProvider(inImage));
	UInt8 * m_PixelBuf = (UInt8 *) CFDataGetBytePtr(m_DataRef);
	UInt8 * m_OutPixelBuf = (UInt8 *) CFDataGetBytePtr(m_OutDataRef);  

	int h = CGImageGetHeight(inImage);
	int w = CGImageGetWidth(inImage);

	int kh = [kernel count] / 2;
	int kw = [[kernel objectAtIndex:0] count] / 2;
	int i = 0, j = 0, n = 0, m = 0;

	for (i = 0; i < h; i++) {
		for (j = 0; j < w; j++) {
			int outIndex = (i*w*4) + (j*4);
			double r = 0, g = 0, b = 0;
			for (n = -kh; n <= kh; n++) {
				for (m = -kw; m <= kw; m++) {
					if (i + n >= 0 && i + n < h) {
						if (j + m >= 0 && j + m < w) {
							double f = [[[kernel objectAtIndex:(n + kh)] objectAtIndex:(m + kw)] doubleValue];
							if (f == 0) {continue;}
							int inIndex = ((i+n)*w*4) + ((j+m)*4);
							r += m_PixelBuf[inIndex] * f;
							g += m_PixelBuf[inIndex + 1] * f;
							b += m_PixelBuf[inIndex + 2] * f;
						}
					}
				}
			}
			m_OutPixelBuf[outIndex]     = SAFECOLOR((int)r);
			m_OutPixelBuf[outIndex + 1] = SAFECOLOR((int)g);
			m_OutPixelBuf[outIndex + 2] = SAFECOLOR((int)b);
			m_OutPixelBuf[outIndex + 3] = 255;
		}
	}

	CGContextRef ctx = CGBitmapContextCreate(m_OutPixelBuf,
											 CGImageGetWidth(inImage),
											 CGImageGetHeight(inImage),
											 CGImageGetBitsPerComponent(inImage),
											 CGImageGetBytesPerRow(inImage),
											 CGImageGetColorSpace(inImage),
											 CGImageGetBitmapInfo(inImage)
											 );

	CGImageRef imageRef = CGBitmapContextCreateImage(ctx);
	CGContextRelease(ctx);
	UIImage *finalImage = [UIImage imageWithCGImage:imageRef];
	CGImageRelease(imageRef);
	CFRelease(m_DataRef);
    CFRelease(m_OutDataRef);

	return finalImage;
}

方法的参数kernel是卷积运算中的卷积核，下面是几种滤镜的卷积核：

#pragma mark -
#pragma mark - Basic Convolutions

/* Reference :
 * http://docs.gimp.org/en/plug-in-convmatrix.html
 */

- (UIImage *)sharpen
{
//	double dKernel[5][5] = {
//		{0,  0.0, -1.0,  0.0, 0},
//		{0, -1.0,  5.0, -1.0, 0},
//		{0,  0.0, -1.0,  0.0, 0}
//    };

    double dKernel[5][5] = {
		{0, 0.0, -0.2,  0.0, 0},
		{0, -0.2, 1.8, -0.2, 0},
		{0, 0.0, -0.2,  0.0, 0}
    };

	NSMutableArray *kernel = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
	for (int i = 0; i < 5; i++) {
		NSMutableArray *row = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
		for (int j = 0; j < 5; j++) {
			[row addObject:[NSNumber numberWithDouble:dKernel[i][j]]];
		}
		[kernel addObject:row];
	}
	return [self applyConvolution:kernel];
}

- (UIImage *)edgeEnhance
{
	double dKernel[5][5] = {
		{0,  0.0,  0.0,  0.0, 0},
		{0, -1.0,  1.0,  0.0, 0},
		{0,  0.0,  0.0,  0.0, 0}
    };

	NSMutableArray *kernel = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
	for (int i = 0; i < 5; i++) {
		NSMutableArray *row = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
		for (int j = 0; j < 5; j++) {
			[row addObject:[NSNumber numberWithDouble:dKernel[i][j]]];
		}
		[kernel addObject:row];
	}

	return [self applyConvolution:kernel];
}

- (UIImage *)edgeDetect
{
	double dKernel[5][5] = {
		{0,  0.0,  1.0,  0.0, 0},
		{0,  1.0, -4.0,  1.0, 0},
		{0,  0.0,  1.0,  0.0, 0}
    };

	NSMutableArray *kernel = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
	for (int i = 0; i < 5; i++) {
		NSMutableArray *row = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
		for (int j = 0; j < 5; j++) {
			[row addObject:[NSNumber numberWithDouble:dKernel[i][j]]];
		}
		[kernel addObject:row];
	}

	return [self applyConvolution:kernel];
}

- (UIImage *)emboss
{
	double dKernel[5][5] = {
		{0, -2.0, -1.0,  0.0, 0},
		{0, -1.0,  1.0,  1.0, 0},
		{0,  0.0,  1.0,  2.0, 0}
    };

	NSMutableArray *kernel = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
	for (int i = 0; i < 5; i++) {
		NSMutableArray *row = [[[NSMutableArray alloc] initWithCapacity:5] autorelease];
		for (int j = 0; j < 5; j++) {
			[row addObject:[NSNumber numberWithDouble:dKernel[i][j]]];
		}
		[kernel addObject:row];
	}

	return [self applyConvolution:kernel];
}

在此基础上，我Google了下Photoshop中对照片进行黑白处理的简单步骤：

1. 去色
2. 调整对比度
3. 高斯模糊
4. 浮雕效果
5. 边缘检测
6. 调整对比度
7. 调整亮度
8. 反相

我按步骤实现了相应代码：

return [[[[[[[[originImage desaturate]
                          changeContrastByFactor:1.5]
                         gaussianBlur:1.3] emboss]
                       edgeDetect]
                      changeContrastByFactor:1.5]
                     changeBrightnessByFactor:1.5]
                    invert];

可惜效果有点粗糙，照片仍旧以上一篇文章中的Andy为例：