Other Projects

I use my spare time to realize and reproduce many classical image processing algorithms. Most of the inspiration for these algorithms comes from the course reports of other students in our course <Human Perception and Electronic Media>.

I was interested in the technology they showed that I decided to make most of the programs I was interested in and start self-learning. The implementation of each of these works did not take so long time. The final works are shown below.

Image restoration technology has a high commercial value at present. There are a group of people in the current market who are engaged in repairing old photos, and I have basically understand the methods they use. I use Python to repair the damaged image and the watermarked image by using the hydrodynamic method and the anti-color neutralization method respectively, and have achieved good results.

The following examples are:

Original vs damaged vs inpainting

The next example is watermarking removal:

Notice the watermarking "会小二" at the bottom of the image on the left. And it is removed in the right image.

Due to the different focal lengths of the camera, blurring will inevitably occur when the photograph is taken near the object.

I use Matlab to design a depth-of-field synthesis program, which can synthesize pictures taken with different focal lengths, and get very clear close-range pictures. This technology is necessary for the expansion of camera functions.

Here is an example:

When the camera takes a close-range object, different parts of the object will be blurred due to different focal lengths.

After depth of field synthesis, we can get a very clear picture.