I accidentally stumbled upon this channel and wondering "WHERE WAS IT ALL THIS TIME?"...I'm going to start from video 1 all the way up to learn OPEN CV
How could bluring is denoising? I am learning image processing for computer vision and it is also said that blurring (spesifically gaussian blurring/smoothing) is denoising algorithm? It even makes information inside the image less clear
Good question. The answer is ... denoising is a side effect of gaussian blurring. When you select small kernel sizes the resulting image may be useful with minimal loss in information. But for large kernel sizes you may lose information due to blurring. Therefore Gaussian is usually not preferred for denoising if edge preservation or texture information is useful.
Suppose we have two images... One original and another one is blurr.. Now we have to overlap these two images with DCT transform then take averaging of d sample.... Then take idct..
There is no simple filter I know of for haze removal. Please search for dark channel prior for haze removal, it is an old classical approach but works fine. I do not have code ready to share for that. I think neural network approach may work better. I will have to experiment and if it works I will create a video and share the code. Here is a good reference: openaccess.thecvf.com/content_ICCV_2017/papers/Li_AOD-Net_All-In-One_Dehazing_ICCV_2017_paper.pdf
I accidentally stumbled upon this channel and wondering "WHERE WAS IT ALL THIS TIME?"...I'm going to start from video 1 all the way up to learn OPEN CV
Thank.u sir ALLAH almighty bless u amin
thank you very much sir
Do you have a video on structured light projected color bands to denoise them and deal with color bleed? Thx.
How could bluring is denoising? I am learning image processing for computer vision and it is also said that blurring (spesifically gaussian blurring/smoothing) is denoising algorithm? It even makes information inside the image less clear
Good question. The answer is ... denoising is a side effect of gaussian blurring. When you select small kernel sizes the resulting image may be useful with minimal loss in information. But for large kernel sizes you may lose information due to blurring. Therefore Gaussian is usually not preferred for denoising if edge preservation or texture information is useful.
@@DigitalSreeni what do you think a good denoising method if I want to preserve edge and remove other noises?
@@DigitalSreeni it can be applied for fundus photography?
Thank you for your work
You are very welcome
at 3:35, how do we apply that matrix to the image?
Please watch full video for answer to your question.
Is itbpossable to unblur an image
Yes, many ways. Please look up deep learning based deblurring.
Great
Suppose we have two images... One original and another one is blurr.. Now we have to overlap these two images with DCT transform then take averaging of d sample.... Then take idct..
How to Remove Haze from images??
There is no simple filter I know of for haze removal. Please search for dark channel prior for haze removal, it is an old classical approach but works fine. I do not have code ready to share for that. I think neural network approach may work better. I will have to experiment and if it works I will create a video and share the code. Here is a good reference: openaccess.thecvf.com/content_ICCV_2017/papers/Li_AOD-Net_All-In-One_Dehazing_ICCV_2017_paper.pdf
try increasing contrast.
Why the image size of gaussian is increase from real size of image before filtering gaussian?
PLEASE HELP ME
Help with what?