基于拉普拉斯算子的模糊图像评价

最新推荐文章于 2025-03-11 23:13:44 发布

NotEmptyLab

最新推荐文章于 2025-03-11 23:13:44 发布

阅读量4.2k

点赞数 3

分类专栏： Python 文章标签： opencv laplacian

本文链接：https://blog.youkuaiyun.com/shawzg/article/details/103729567

版权

Python 专栏收录该内容

4 篇文章

订阅专栏

一. 模糊图像评价基础知识

图像质量评价(IQA)，根据参考图片(reference image)，即原始图片的存在与否，可分为：

全参考（full-reference）方法, 有原始图片的全部信息
半参考（reduced-reference）方法, 只有原始图片的部分信息
无参考（no-reference）方法，没有原始图片

模糊图片分类

defocus blur image 散焦模糊图像
motion blur image 运动模糊图像

图像模糊原因:

确定性因素：包括成像系统调焦不当、目标物体相对运动等
随机性因素：主要是图像在记录、传输过程中的污染，比如电子系统的高频性能不好造成图像高频分量的损失。

在频域，当一幅图像的高频部分被削弱时图像看起来会显得模糊。

在空域，图像的边界和细节部分不清晰时，图像看起来也会显得模糊。

二. 无参考图片算法实现(Python)

import cv2
import os
import time

class Laplacian :
    def __init__(self, image_dir, recursion = False) :
        self.__image_dir = image_dir
        self.__image_paths = []
        for root_dir, sub_dirs, images in os.walk(image_dir, recursion) :
            for image in images :
                self.__image_paths.append(os.path.join(root_dir, image))

    def __read_image(self, image_path) :
        # In the case of color images, the decoded images will have the channels stored in **B G R** order.
        # When using IMREAD_GRAYSCALE, the codec's internal grayscale conversion will be used, if available.
        # Results may differ to the output of cvtColor()
        return cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)

    def __pre_process_image(self, image_path, **kwargs) :
        image = self.__read_image(image_path)
        if kwargs.get('width') and kwargs.get('height') :
            image = cv2.resize(image, (kwargs.get('width'), kwargs.get('height')))
        return image

    def get_image_paths(self) :
        return self.__image_paths

    def blur_detect(self, image_path) :
        cv_image = self.__pre_process_image(image_path, width = 200, height = 200)
        return cv2.Laplacian(cv_image, cv2.CV_64F).var()


def test_laplacian() :
    laplacian = Laplacian('dir of images that you want to detect with laplacian', False)
    image_paths = laplacian.get_image_paths()
    for path in image_paths :
        begin = time.time() * 1000
        score = laplacian.blur_detect(path)
        end = time.time() * 1000
        print("Laplacian detection {} value is {}, cost time {} ms".format(path, int(score), end - begin))


if __name__ == "__main__" :
    test_laplacian()

代码核心是OpenCV提供的接口cv2.Laplacian，

def Laplacian(src, ddepth, dst=None, ksize=None, scale=None, delta=None, borderType=None):
“”"
Laplacian(src, ddepth[, dst[, ksize[, scale[, delta[, borderType]]]]]) -> dst
. @brief Calculates the Laplacian of an image.
.
. The function calculates the Laplacian of the source image by adding up the second x and y
. derivatives calculated using the Sobel operator:
.
. \f[\texttt{dst} = \Delta \texttt{src} = \frac{\partial^2 \texttt{src}}{\partial x^2} + \frac{\partial^2 \texttt{src}}{\partial y^2}\f]
.
. This is done when ksize > 1. When ksize == 1, the Laplacian is computed by filtering the image
. with the following \f $KaTeX parse error: Undefined control sequence: \f at position 11: 3 \times 3\̲f̲$ aperture:
.
. \f[\vecthreethree {0}{1}{0}{1}{-4}{1}{0}{1}{0}\f]
.
. @param src Source image.
. @param dst Destination image of the same size and the same number of channels as src .
. @param ddepth Desired depth of the destination image.
. @param ksize Aperture size used to compute the second-derivative filters. See #getDerivKernels for
. details. The size must be positive and odd.
. @param scale Optional scale factor for the computed Laplacian values. By default, no scaling is
. applied. See #getDerivKernels for details.
. @param delta Optional delta value that is added to the results prior to storing them in dst .
. @param borderType Pixel extrapolation method, see #BorderTypes
. @sa Sobel, Scharr
“”"
pass