[200 opencv routines] 220 Mosaic the image

List of articles ：『youcans Of OpenCV routine 200 piece - General catalogue 』

【youcans Of OpenCV routine 200 piece 】220. Mosaic the image

9. Mosaic processing of image

Mosaic effect is a widely used image and video processing method . Degrade the color scale details of the specified area in the image , Cause the effect of color block blur , It looks like a color block composed of small squares , Called mosaic . The main purpose of mosaic effect is to make the details of specific areas unrecognizable , It is often used to cover people's faces 、 Privacy information .

The mosaic method is very simple , Divide the processing area into small blocks , Set all pixels in each small square to the same or similar pixel values . routine A4.13 A simple implementation case is given .

The larger the size of the mosaic square , The more blurred the image , The more details the mosaic area image loses .

This is similar to image multiscale pixel sampling ： Image down sampling , The resolution decreases step by step . Mosaic the whole image , It is equivalent to image down sampling ; Mosaic the part of the image , It is equivalent to the fusion of down sampling of the original image and local image .

Up sampling and down sampling are irreversible , High frequency information will be lost when the down sampled image is restored to its original size , Blur the picture . therefore , It is impossible in principle to eliminate image mosaic . however , Filling algorithm through interpolation of adjacent points of picture pixels , It can enhance the visual effect of mosaic area . In recent years , With AI Technological development , Based on learning a large number of similar clear images , Use AI The algorithm can restore the image better , Recognize faces or texts blocked by mosaics , Good results were achieved .

routine A4.13： Mosaic the specified area of the image

For the selected ROI Mosaic the area . The larger the size of the mosaic square , The more blurred the image , The more details the mosaic area image loses .

    # A4.13  Mosaic the specified area of the image 
    img = cv.imread("../images/imgLena.tif", 1)  #  Load the original image , single channel 
    roi = cv.selectROI(img, showCrosshair=True, fromCenter=False)
    x, y, wRoi, hRoi = roi  #  Position parameters of rectangular clipping region 
    # x, y, wRoi, hRoi = 208, 176, 155, 215 #  Rectangular crop region 
    imgROI = img[y:y+hRoi, x:x+wRoi].copy()  #  Slice to get a rectangular clipping region 
    print(x, y, wRoi, hRoi)

    plt.figure(figsize=(9, 6))
    plt.subplot(231), plt.title("Original image"), plt.axis('off')
    plt.imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))
    plt.subplot(232), plt.title("Region of interest"), plt.axis('off')
    plt.imshow(cv.cvtColor(imgROI, cv.COLOR_BGR2RGB))

    mosaic = np.zeros(imgROI.shape, np.uint8)  # ROI  Area 
    ksize = [5, 10, 20]  #  Width of mosaic block 
    for i in range(3):
        k = ksize[i]
        for h in range(0, hRoi, k):
            for w in range(0, wRoi, k):
                color = imgROI[h,w]
                mosaic[h:h+k,w:w+k,:] = color  #  Cover the mosaic block with vertex color 
        imgMosaic = img.copy()
        imgMosaic[y:y + hRoi, x:x + wRoi] = mosaic
        plt.subplot(2,3,i+4), plt.title("Coding image (size={})".format(k)), plt.axis('off')
        plt.imshow(cv.cvtColor(imgMosaic, cv.COLOR_BGR2RGB))

    plt.subplot(233), plt.title("Mosaic"), plt.axis('off')
    plt.imshow(cv.cvtColor(mosaic, cv.COLOR_BGR2RGB))
    plt.show()

【 At the end of this section 】

Copyright notice ：
[email protected] Original works , Reprint must be marked with the original link ：(https://blog.csdn.net/youcans/article/details/125522759)
Copyright 2022 youcans, XUPT
Crated：2022-6-30

218. Multi line oblique text watermark
219. Add digital blind watermark
220.220. Mosaic the image