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：

[cpp]
- (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：
[cpp]
#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：
去色
调整对比度
高斯模糊
浮雕效果
边缘检测
调整对比度
调整亮度
反相
我按步骤实现了相应代码：

[cpp] 
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：