• iOS图片相似度比较


    1.      缩小尺寸:将图像缩小到8*8的尺寸,总共64个像素。这一步的作用是去除图像的细节,只保留结构/明暗等基本信息,摒弃不同尺寸/比例带来的图像差异;

    注:实际操作时,采取了两种尺寸作对比(10*10,100*100)尺寸再大的话就会性能影响就会较大了,我实现了两种,目的是为了展示怎么设定不同尺寸。

    2.      简化色彩:将缩小后的图像,转为64级灰度,即所有像素点总共只有64种颜色;

    注:关于多少灰度级的问题,我并没有太在意,采取了一个合适的RGB to GRAY 算法就好,个人理解

    3.      计算平均值:计算所有64个像素的灰度平均值;

    4.      比较像素的灰度:将每个像素的灰度,与平均值进行比较,大于或等于平均值记为1,小于平均值记为0;

    5.      计算哈希值:将上一步的比较结果,组合在一起,就构成了一个64位的整数,这就是这张图像的指纹。组合的次序并不重要,只要保证所有图像都采用同样次序就行了;

    6.      得到指纹以后,就可以对比不同的图像

    GetSimilarity.h

    //
    //  GetSimilarity.h
    //  imgsimlartest
    //
    //  Created by test on 16/3/3.
    //  Copyright © 2016年 com.facishare.CoreTest. All rights reserved.
    //
    
    #import <Foundation/Foundation.h>
    #import <UIKit/UIKit.h>
    typedef double Similarity;
    
    @interface GetSimilarity : NSObject
    - (void)setImgWithImgA:(UIImage*)imgA ImgB:(UIImage*)imgB;//设置需要对比的图片
    - (void)setImgAWidthImg:(UIImage*)img;
    - (void)setImgBWidthImg:(UIImage*)img;
    - (Similarity)getSimilarityValue; //获取相似度
    + (Similarity)getSimilarityValueWithImgA:(UIImage*)imga ImgB:(UIImage*)imgb;//类方法
    @end
    

    GetSimilarity.m

    //
    //  GetSimilarity.m
    //  imgsimlartest
    //
    //  Created by test on 16/3/3.
    //  Copyright © 2016年 com.facishare.CoreTest. All rights reserved.
    //
    
    #import "GetSimilarity.h"
    #define ImgSizeA 10
    #define ImgSizeB 100
    typedef enum workday
    {
        SizeA,
        SizeB,
    }GetSimilarityType;
    
    
    @interface GetSimilarity()
    @property (nonatomic,assign) Similarity similarity;
    @property (nonatomic,strong) UIImage *imga;
    @property (nonatomic,strong) UIImage *imgb;
    @end
    
    @implementation GetSimilarity
    - (instancetype)init
    {
        self = [super init];
        if (self) {
            self.imga = [[UIImage alloc]init];
            self.imgb = [[UIImage alloc]init];
        }
        return self;
    }
    
    - (void)setImgWithImgA:(UIImage*)imgA ImgB:(UIImage*)imgB
    {
        _imga = imgA;
        _imgb = imgB;
    }
    
    - (void)setImgAWidthImg:(UIImage*)img
    {
        self.imga = img;
    }
    
    - (void)setImgBWidthImg:(UIImage*)img
    {
        self.imgb = img;
    }
    
    - (Similarity)getSimilarityValue
    {
        self.similarity = MAX([self getSimilarityValueWithType:SizeA], [self getSimilarityValueWithType:SizeB]);
        return self.similarity;
    } 
    + (Similarity)getSimilarityValueWithImgA:(UIImage *)imga ImgB:(UIImage *)imgb
    {
        GetSimilarity * getSimilarity = [[GetSimilarity alloc]init];
        [getSimilarity setImgWithImgA:imga ImgB:imgb];
        return [getSimilarity getSimilarityValue];
    }
    - (Similarity)getSimilarityValueWithType:(GetSimilarityType)type;//
    {
        int cursize = (type == SizeA ? ImgSizeA : ImgSizeB);
        int ArrSize = cursize * cursize + 1,a[ArrSize],b[ArrSize],i,j,grey,sum = 0;
        CGSize size = {cursize,cursize};
        UIImage * imga = [self reSizeImage:self.imga toSize:size];
        UIImage * imgb = [self reSizeImage:self.imgb toSize:size];//缩小图片尺寸
    
        a[ArrSize] = 0;
        b[ArrSize] = 0;
        CGPoint point;
        for (i = 0 ; i < cursize; i++) {//计算a的灰度
            for (j = 0; j < cursize; j++) {
                point.x = i;
                point.y = j;
                grey = ToGrey([self UIcolorToRGB:[self colorAtPixel:point img:imga]]);
                a[cursize * i + j] = grey;
                a[ArrSize] += grey;
            }
        }
        a[ArrSize] /= (ArrSize - 1);//灰度平均值
        for (i = 0 ; i < cursize; i++) {//计算b的灰度
            for (j = 0; j < cursize; j++) {
                point.x = i;
                point.y = j;
                grey = ToGrey([self UIcolorToRGB:[self colorAtPixel:point img:imgb]]);
                b[cursize * i + j] = grey;
                b[ArrSize] += grey;
            }
        }
        b[ArrSize] /= (ArrSize - 1);//灰度平均值
        for (i = 0 ; i < ArrSize ; i++)//灰度分布计算
        {
            a[i] = (a[i] < a[ArrSize] ? 0 : 1);
            b[i] = (b[i] < b[ArrSize] ? 0 : 1);
        }
        ArrSize -= 1;
        for (i = 0 ; i < ArrSize ; i++)
        {
            sum += (a[i] == b[i] ? 1 : 0);
        }
        
        return sum * 1.0 / ArrSize;
    }
    
    - (UIImage *)reSizeImage:(UIImage *)image toSize:(CGSize)reSize//重新设定图片尺寸
    {
        UIGraphicsBeginImageContext(CGSizeMake(reSize.width, reSize.height));
        [image drawInRect:CGRectMake(0, 0, reSize.width, reSize.height)];
        UIImage *reSizeImage = UIGraphicsGetImageFromCurrentImageContext();
        UIGraphicsEndImageContext();
        return reSizeImage;
    }
    
    unsigned int ToGrey(unsigned int rgb)//RGB计算灰度
    {
        unsigned int blue   = (rgb & 0x000000FF) >> 0;
        unsigned int green  = (rgb & 0x0000FF00) >> 8;
        unsigned int red    = (rgb & 0x00FF0000) >> 16;
        return ( red*38 +  green * 75 +  blue * 15 )>>7;
    }
    
    - (unsigned int)UIcolorToRGB:(UIColor*)color//UIColor转16进制RGB
    {
        unsigned int RGB,R,G,B;
        RGB = R = G = B = 0x00000000;
        CGFloat r,g,b,a;
        [color getRed:&r green:&g blue:&b alpha:&a];
        R = r * 256 ;
        G = g * 256 ;
        B = b * 256 ;
        RGB = (R << 16) | (G << 8) | B ;
        return RGB;
    }
    
    - (UIColor *)colorAtPixel:(CGPoint)point img:(UIImage*)img{//获取指定point位置的RGB
        // Cancel if point is outside image coordinates
        if (!CGRectContainsPoint(CGRectMake(0.0f, 0.0f, img.size.width, img.size.height), point)) { return nil; }
        
        NSInteger   pointX  = trunc(point.x);
        NSInteger   pointY  = trunc(point.y);
        CGImageRef  cgImage = img.CGImage;
        NSUInteger  width   = img.size.width;
        NSUInteger  height  = img.size.height;
        int bytesPerPixel   = 4;
        int bytesPerRow     = bytesPerPixel * 1;
        NSUInteger bitsPerComponent = 8;
        unsigned char pixelData[4] = { 0, 0, 0, 0 };
        CGColorSpaceRef colorSpace = CGColorSpaceCreateDeviceRGB();
        CGContextRef context = CGBitmapContextCreate(pixelData, 1, 1, bitsPerComponent, bytesPerRow, colorSpace, kCGImageAlphaPremultipliedLast | kCGBitmapByteOrder32Big);
        
        CGColorSpaceRelease(colorSpace);
        CGContextSetBlendMode(context, kCGBlendModeCopy);
        
        // Draw the pixel we are interested in onto the bitmap context
        CGContextTranslateCTM(context, -pointX, pointY-(CGFloat)height);
        CGContextDrawImage(context, CGRectMake(0.0f, 0.0f, (CGFloat)width, (CGFloat)height), cgImage);
        CGContextRelease(context);
        // Convert color values [0..255] to floats [0.0..1.0]
      
        CGFloat red   = (CGFloat)pixelData[0] / 255.0f;
        CGFloat green = (CGFloat)pixelData[1] / 255.0f;
        CGFloat blue  = (CGFloat)pixelData[2] / 255.0f;
        CGFloat alpha = (CGFloat)pixelData[3] / 255.0f;
        return [UIColor colorWithRed:red green:green blue:blue alpha:alpha];
    }
    @end
    

      

  • 相关阅读:
    LeetCode——230. 二叉搜索树中第K小的元素
    LeetCode——456.132模式
    LeetCode——623.在二叉树中增加一行
    LeetCode——735.行星碰撞
    翻译——2_Linear Regression and Support Vector Regression
    LeetCode——919.完全二叉树插入器
    论文翻译——Deep contextualized word representations
    LeetCode——853.车队
    Leetcode——863.二叉树中所有距离为 K 的结点
    Pytorch——BERT 预训练模型及文本分类
  • 原文地址:https://www.cnblogs.com/yangyuxiaozi/p/7419349.html
Copyright © 2020-2023  润新知