媒体大数据 视频分镜与镜头相似度曲线

四个计算相似度的算法

#四种检测方法
import os
import math
import cv2
import numpy as np
import matplotlib.pyplot as plt
#均值哈希
os.chdir("F:\\视频\\")
#均值哈希
def aHash(img):    # 缩放为8*8
    img = cv2.resize(img, (8, 8))
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)#归一化为灰度图
    # s为像素和初值为0，hash_str为hash值初值为''
    s = 0
    hash_str = ''    # 遍历累加求像素和
    for i in range(8):
        for j in range(8):
            s = s + gray[i, j]    # 求平均灰度
    avg = s / 64    # 灰度大于平均值为1相反为0生成图片的hash值
    for i in range(8):
        for j in range(8):
            if gray[i, j] > avg:
                hash_str = hash_str + '1'
            else:
                hash_str = hash_str + '0'
    return hash_str
def cmphash(hash1,hash2):
    n=0
    if len(hash1)!=len(hash2):
        return -1
    for i in range(len(hash1)):
        if hash1[i]!=hash2[i]:
            n=n+1;
    n=n/len(hash1)
    return n
#差值哈希算法
def dHash(img):
    # 缩放8*8
    img = cv2.resize(img, (9, 8))
    # 转换灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    img = cv2.resize(img, (32, 32))
    hash_str = ''
    # 每行前一个像素大于后一个像素为1，相反为0，生成哈希
    for i in range(8):
        for j in range(8):
            if gray[i, j] > gray[i, j + 1]:
                hash_str = hash_str + '1'
            else:
                hash_str = hash_str + '0'

    return hash_str
#感知哈希
def pHash(img):
    if img is None:
        print("Image is empty.")
    else:
        img=cv2.resize(img,(32,32))
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
        dct=cv2.dct(np.float32(gray))
        dct_roi=dct[0:8,0:8]
        hash=[]
        average=np.mean(dct_roi)
        for i in range(dct_roi.shape[0]):
            for j in range(dct_roi.shape[1]):
                if dct_roi[i,j]>average:
                    hash.append(1)
                else:
                    hash.append(0)
        return hash
#RGB直方图相似度
def calculate(image1, image2):
    hist1 = cv2.calcHist([image1], [0], None, [256], [0.0, 255.0])
    hist2 = cv2.calcHist([image2], [0], None, [256], [0.0, 255.0])
    degree = 0
    for i in range(len(hist1)):
        if hist1[i] != hist2[i]:
            degree = degree + (1 - abs(hist1[i] - hist2[i]) / max(hist1[i], hist2[i]))
        else:
            degree = degree + 1
    degree = degree / len(hist1)
    return degree
def classify_hist_with_split(image1, image2, size=(256, 256)):
    # 将图像resize后，分离为RGB三个通道，再计算每个通道的相似值
    image1 = cv2.resize(image1, size)
    image2 = cv2.resize(image2, size)
    sub_image1 = cv2.split(image1)
    sub_image2 = cv2.split(image2)
    sub_data = 0
    for im1, im2 in zip(sub_image1, sub_image2):
        sub_data += calculate(im1, im2)
    sub_data = sub_data / 3
    return sub_data
#四种调用

视频分镜处理

def test2(thresholds,ways):
    os.chdir("C:\\Users\\33361\\PycharmProjects\\F1\\")
    v_path = 'static/巴斯特斯克鲁格斯的歌谣决斗场面.mp4'
    cap=cv2.VideoCapture(v_path)
    fc=cap.get(cv2.CAP_PROP_FRAME_COUNT)
    fps=cap.get(cv2.CAP_PROP_FPS)
    scene_changes = []
    similarities = []
    print(fc)
    _, img1 = cap.read()
    cv2.imwrite('static/{}/{}/image{}.jpg'.format(ways,str(thresholds),0), img1)
    print(int(fc))
    for i in range(1,int(fc)):
        _, img2 = cap.read()
        if ways=='ahash':
            hash1=aHash(img1)
            hash2=aHash(img2)
            n=cmphash(hash1,hash2)
            similarities.append(1-n)
            if (1-n < thresholds):
                cv2.imwrite('static/{}/{}/image{}.jpg'.format(ways, str(thresholds), i), img2)
                scene_changes.append(i)
                img1 = img2
        if ways == 'phash':
            hash1 = pHash(img1)
            hash2 = pHash(img2)
            n = cmphash(hash1, hash2)
            similarities.append(1-n)
            if (1-n < thresholds):
                cv2.imwrite('static/{}/{}/image{}.jpg'.format(ways, str(thresholds), i), img2)
                scene_changes.append(i)
                img1 = img2
        if ways == 'dhash':
            hash1 = dHash(img1)
            hash2 = dHash(img2)
            n = cmphash(hash1, hash2)
            similarities.append(1-n)
            if (1-n < thresholds):
                cv2.imwrite('static/{}/{}/image{}.jpg'.format(ways, str(thresholds), i), img2)
                scene_changes.append(i)
                img1 = img2
        if ways == 'hist':
            n=classify_hist_with_split(img1, img2)
            similarities.append(n)
            if (n < thresholds):
                cv2.imwrite('static/{}/{}/image{}.jpg'.format(ways, str(thresholds), i), img2)
                scene_changes.append(i)
                img1 = img2
    return scene_changes,similarities,fps

绘画相似度曲线

def calculate_accuracy(similarities, scene_changes, threshold, ways, fps):
    # 处理场景变更列表，将相邻的变更点合并成一个区间
    fig, ax = plt.subplots(figsize=(9, 6))
    ax.plot(np.arange(len(similarities)) / fps, similarities)
    ax.set_xlabel('时间/s')
    ax.set_ylabel('相似度')
    ax.set_title(f"算法为{ways},阈值为{threshold}时的相似度曲线图")
    for m in range(len(similarities)):
        if ways == "hist":
             if  similarities[m] < threshold:    
                ax.scatter(m/fps, similarities[m], color='red', marker='o')
        else:
             if  similarities[m] < threshold:    
                ax.scatter(m/fps, similarities[m], color='red', marker='o')
    # 绘制相似度曲线，并在分镜位置做标记
    plt.savefig("{}算法下阈值{}相似度曲线图.png".format(ways,threshold))  # 保存图片到文件
    plt.show()

主函数

if __name__ == '__main__':
    import matplotlib
    matplotlib.rc("font", family='YouYuan')
    filename = '巴斯特斯克鲁格斯的歌谣决斗场面.mp4'
    way=["phash","ahash","dhash","hist"]
    scene_changes= []
    accury=0
    for i in range(len(way)):
        scene_changes, similarities,fps = test2(0.6,way[i])
        calculate_accuracy(similarities, scene_changes, 0.6, way[i],fps)
        

曲线效果