传入一张图，生成它的油画版！（python实现）

迈不过友情╰ 2023-07-03 16:24 220阅读 0赞

### 文章目录 ###

*   *  一、函数
     *   *  1、主函数
         *  2、使用方法
     *  二、效果展示
     *   *  1、原图像
         *  2、油画版效果

## 一、函数 ##

### 1、主函数 ###

import cv2
    import numpy
    
    
    def oilPainting(img, templateSize, bucketSize, step):  # templateSize模板大小,bucketSize桶阵列,step模板滑动步长
    
        gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
        gray = ((gray / 256) * bucketSize).astype(int)  # 灰度图在桶中的所属分区
        h, w = img.shape[:2]
    
        oilImg = numpy.zeros(img.shape, numpy.uint8)  # 用来存放过滤图像
    
        for i in range(0, h, step):
    
            top = i - templateSize
            bottom = i + templateSize + 1
            if top < 0:
                top = 0
            if bottom >= h:
                bottom = h - 1
    
            for j in range(0, w, step):
    
                left = j - templateSize
                right = j + templateSize + 1
                if left < 0:
                    left = 0
                if right >= w:
                    right = w - 1
    
                # 灰度等级统计
                buckets = numpy.zeros(bucketSize, numpy.uint8)  # 桶阵列，统计在各个桶中的灰度个数
                bucketsMean = [0, 0, 0]  # 对像素最多的桶，求其桶中所有像素的三通道颜色均值
                # 对模板进行遍历
                for c in range(top, bottom):
                    for r in range(left, right):
                        buckets[gray[c, r]] += 1  # 模板内的像素依次投入到相应的桶中，有点像灰度直方图
    
                maxBucket = numpy.max(buckets)  # 找出像素最多的桶以及它的索引
                maxBucketIndex = numpy.argmax(buckets)
    
                for c in range(top, bottom):
                    for r in range(left, right):
                        if gray[c, r] == maxBucketIndex:
                            bucketsMean += img[c, r]
                bucketsMean = (bucketsMean / maxBucket).astype(int)  # 三通道颜色均值
    
                # 油画图
                for m in range(step):
                    for n in range(step):
                        oilImg[m + i, n + j] = (bucketsMean[0], bucketsMean[1], bucketsMean[2])
        return oilImg
    
    
    img = cv2.imread(r'my.jpg', cv2.IMREAD_ANYCOLOR)
    oil = oilPainting(img, 4, 8, 2)
    cv2.imshow('youhua', oil)
    cv2.imwrite(r'myyouhua.jpg', oil)

### 2、使用方法 ###

*  安装python相关环境，更改其图片路径和图片名即可，计算时间较长，耐心等待。

## 二、效果展示 ##

### 1、原图像 ###

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ_size_16_color_FFFFFF_t_70]

### 2、油画版效果 ###

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ_size_16_color_FFFFFF_t_70 1]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20200207152541144.jpg?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ==,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20200207152421596.jpg?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L21pcmFjbGVvYQ==,size_16,color_FFFFFF,t_70