学 士 学 位 论 文
BACHELOR ’S THESIS
Abstract
This paper presents a Python-based image dehazing algorithm that relies primarily
on two techniques: the dark channel prior and inverse depth estimation. These two
techniques are currently the hot research directions in the field of image dehazing and
are used by most mainstream dehazing algorithms.
The dark channel prior is a natural image-based prior knowledge that finds the
global minimum in an image, i.e., the dark channel, and uses this global information
to remove haze. It assumes that any pair of similar regions in a natural image has a
similar minimum value on the dark channel. By calculating the dark channel, the
depth information of the image can be obtained, and then dehazing can be performed.
The dark channel prior is widely used in the field of dehazing and has become the
foundation of many dehazing algorithms.
Inverse depth estimation is a technique that calculates the depth information by
computing the gradient of each pixel in an image, one can ascertain the value of each
pixel depth direction based on the physical principles of the camera optical system. It
is an algorithm based on physical principles and can accurately estimate the depth
information in the image, thereby enabling better dehazing.
As computer hardware and algorithms continue to develop, the field of image
dehazing is also advancing. The directions of their future research are the deep
learn-based haze removal alcohol, dehazing algorithms that combine multiple
techniques, and dehazing algorithms that target specific scenarios. These directions
will improve the accuracy, efficiency, and intelligence of dehazing algorithms to meet
the needs of various practical application scenarios.
In summary, this paper introduces a Python-based image dehazing algorithm and
explores the current trends in technology development. With the continuous
development of computer hardware and algorithms, the development prospects of
image dehazing technology are becoming more and more broad. In the future, we can
expect more accurate, efficient, and intelligent dehazing algorithms to emerge to meet
the needs of various practical application scenarios.
Key Words：Python; Image Dehazing Algorithm; Technological Development; Dark
Channel Prior; Inverse Depth Estimation; Deep Learning; Special Scenes; Dehazing
Effect.
