Technology that is common in movies, clearly clarifies obscure images of surveillance cameras. I think that it is impossible for any engineer, but recently I can not deny you. That is image analysis using deep learning. I will analyze the image in a different way from clarifying.
TensorZoom photo expansion software using such image analysis technology.
How to use TensorZoom
I am using TensorZoom, an Android application that implements TensorZoom.
Original image. It is a picture of a fairly small cat.
Enlarged place. I am blurred.
I am using Tensorflow for analysis. Analysis is done part by part, it takes time as it is.
Image after analysis.
Image before analysis.
After analysis. It is a different atmosphere from what I did to sharpen.
Another image. Before analysis.
After analysis.
TensorZoom uses Tensorflow for analysis as described above. Unlike simply sharpness, different results have come out. As this progresses, it may be possible to make low resolution photos equivalent to high resolution, or to restore a human face from blurred surveillance camera images.
TensorZoom is Python software (the source code is released, but the license is not specified).
No comments:
Post a Comment