[python-opencv] PNG 裁切物体

本文主要是介绍[python-opencv] PNG 裁切物体，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！

拿到一组图PNG的图，边缘有点太宽了，需要裁切一下，为了这个需求，简单复习一下基本语法。

1.  读取PNG的4个通道

image = cv.imread(image_path, cv.IMREAD_UNCHANGED)

2. 遍历文件夹获取每个图像文件：

def traverse_folder(folder_path):for root, dirs, files in os.walk(folder_path):for file in files:# 文件的绝对路径file_path = os.path.join(root, file)print(file_path)

3. 找到PNG中物体的ROI：

def findROI(img_path):image = cv.imread(img_path)gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)# 获取图像的宽高通道height, width, channels = image.shape# 进行边缘检测edges = cv.Canny(gray, 50, 255)xmin = width - 1ymin = height - 1xmax = 0ymax = 0for i in range(width):for j in range(height):if edges[j, i] != 0:if xmin > i:xmin = iif xmax < i:xmax = iif ymin > j:ymin = jif ymax < j:ymax = jprint("width : ", xmax - xmin)print("height : ", ymax - ymin)return xmin,xmax,ymin,ymax

4. 遍历文件夹将jpg转PNG

def traverse_folder_convertJPGtoPNG(folder_path, output_path):for root, dirs, files in os.walk(folder_path):for file in files:# 文件的绝对路径file_path = os.path.join(root, file)print(file_path)# 获取没有后缀的文件名filename_without_extension = os.path.splitext(file)[0]print(filename_without_extension)#print(output_path + '/' + filename_without_extension + '.png')jpg_to_png(file_path, output_path + '/' + filename_without_extension + '.png')def jpg_to_png(jpg_path, png_path):img = Image.open(jpg_path)  # 打开jpg文件img.save(png_path, 'PNG')  # 保存为png文件

5. 按照指定ROI图像获取图像区域并保存

def cropbysize_and_save_image(image_path, output_path, outputWidth,outputHeight):image = cv.imread(image_path, cv.IMREAD_UNCHANGED)height, width, channels = image.shapeprint("图像宽度:", width)print("图像高度:", height)print("图像频道:", channels)ymin=math.floor(height*0.5-outputHeight*0.5)ymax=math.ceil(height*0.5+outputHeight*0.5)xmin=math.floor(width*0.5-outputWidth*0.5)xmax=math.ceil(width*0.5+outputWidth*0.5)cropped_image = image[ymin:ymax, xmin:xmax]# 保存为png文件cv.imwrite(output_path, cropped_image, [int(cv.IMWRITE_PNG_STRATEGY_DEFAULT), 9])  

6. 按照问题所在ROI获取图像区域

def crop_and_save_image(image_path, output_path):image = cv.imread(image_path, cv.IMREAD_UNCHANGED)gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)height, width, channels = image.shapeedges = cv.Canny(gray, 50, 255)print(edges)xmin = width - 1ymin = height - 1xmax = 0ymax = 0for i in range(width):for j in range(height):if edges[j,i] != 0:if xmin > i:xmin = iif xmax < i:xmax = iif ymin > j:ymin = jif ymax < j:ymax = jcropped_image = image[ymin-2:ymax+2, xmin-2:xmax+2]cv.imwrite(output_path,cropped_image, [int(cv.IMWRITE_PNG_STRATEGY_DEFAULT), 9]) 

Imwrite参数说明

这篇关于[python-opencv] PNG 裁切物体的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！

---