【DL】图像去噪中的patch size是什么（1篇文章链接和1段代码）

图像去噪中的patch size是什么（1篇文章链接和1段代码）

写在前面，因为自己不做图像相关的研究，所以具体内容也不敢妄加总结，自己感觉貌似就是将一整张图分成一块一块的进行处理，每个块之间可以重叠。但具体大家可以看下面的文章链接和代码示例。

文章链接：Patch-based models and algorithms for image denoising: a comparative review between patch-based images denoising methods for additive noise reduction

代码示例：

patch_size, stride = 40, 10

def gen_patches(file_name):

    # read image
    img = cv2.imread(file_name, 0)  # gray scale
    h, w = img.shape
    scales = [1, 0.9, 0.8, 0.7]
    patches = []

    for s in scales:
        h_scaled, w_scaled = int(h*s),int(w*s)
        img_scaled = cv2.resize(img, (h_scaled,w_scaled), interpolation=cv2.INTER_CUBIC)
        # extract patches
        for i in range(0, h_scaled-patch_size+1, stride):
            for j in range(0, w_scaled-patch_size+1, stride):
                x = img_scaled[i:i+patch_size, j:j+patch_size]
                patches.append(x)
    
    return patches