“Roumen is an extremely intelligent embedded system expert with strong software design/development capability, and is well experienced in solving engineering problems by applying outstanding theoretical knowledge, especially in digital signal processing domain. I enjoyed working with Roumen in delivering solutions to the first Linux Internet Tablet (Nokia 770) in Nokia as well as couple of other challenging research projects. I recommend Roumen as he is well capable in applying himself to deliver the best solution to any problem.”
About
Contributions
-
How does super resolution improve image quality over interpolation?
If we can make a model of how the scene evolves over time in the video then we can use multiple frames to fit that model and as a result reduce the quantisation/sampling noise (increase resolution). It's like watching a low res video and when you pause it the single frame looks a lot worse.
Activity
-
What this means is that we have neither a theoretical nor a practical grasp of how to build the 'robustness' which characterizes all natural…
What this means is that we have neither a theoretical nor a practical grasp of how to build the 'robustness' which characterizes all natural…
Liked by Roumen Popov
-
I wish people would engage their critical thinking skills when talking about AI hype, in this case an overblown conclusion that "autonomous vehicles…
I wish people would engage their critical thinking skills when talking about AI hype, in this case an overblown conclusion that "autonomous vehicles…
Liked by Roumen Popov
-
“Although it can boost individuals’ creativity, it seems to homogenize and flatten our collective output.” This is one conclusion of a new study…
“Although it can boost individuals’ creativity, it seems to homogenize and flatten our collective output.” This is one conclusion of a new study…
Liked by Roumen Popov
Patents
-
Linear transform-based image processing techniques
Issued US US9189833B2
Signal processing techniques utilize operations performed in linear transform domains to perform operations including noise reduction, noise shaping, and gradient integration. The standard wavelet shrinkage method may be altered to constrain the modification of wavelet coefficients towards an initial estimate of the adjusted coefficients. The initial estimates can be computed by applying an edge detection filter to an input image. The wavelet shrinkage method may additionally be altered to…
Signal processing techniques utilize operations performed in linear transform domains to perform operations including noise reduction, noise shaping, and gradient integration. The standard wavelet shrinkage method may be altered to constrain the modification of wavelet coefficients towards an initial estimate of the adjusted coefficients. The initial estimates can be computed by applying an edge detection filter to an input image. The wavelet shrinkage method may additionally be altered to include a noise preservation factor that enables an amount of noise to be preserved to avoid the production of artifacts. Integration of modified multi-dimensional gradients may also be performed in the wavelet (or other linear transform) domain by performing simple integration of the gradient (summing) separately in each dimension, obtaining a linear transform representation of the resulting signals, and combining the linear transforms.
Projects
-
Robbie - an autonomous mobile home robot
-
Robbie is a personal project about building an autonomous mobile home robot capable of doing small useful things at home, such as bringing a beer or a bottle of water from the kitchen, playing with the kids (throwing a ball, bowling, arranging cubes), playing games like physical chess and backgammon, etc.
The robot software architecture consists of a stereo vision module, visual motion estimation
module, navigation module, and control logic module that determines the robot…Robbie is a personal project about building an autonomous mobile home robot capable of doing small useful things at home, such as bringing a beer or a bottle of water from the kitchen, playing with the kids (throwing a ball, bowling, arranging cubes), playing games like physical chess and backgammon, etc.
The robot software architecture consists of a stereo vision module, visual motion estimation
module, navigation module, and control logic module that determines the robot behavior.
The stereo vision module uses a semi-dense stereo vision algorithm that I developed from the
ground up. It is based on three cameras in an L shape configuration to tackle the problem with
horizontal lines. It uses local polynomial approximation of edges and ridges to generate robust
features and to achieve significantly improved sub-pixel accuracy of the disparity estimation,
which allows reducing the frame size considerably to save on computational requirements. The
algorithm was implemented in OpenCL on CPU using the SSE vector engine and fixed point
arithmetic.
The visual motion estimation module uses a semi-dense visual motion estimation algorithm
based on spatial data produced by the stereo vision module. It was also developed from the
ground up. I implemented the algorithm in OpenCL on CPU using the SSE vector engine and
fixed point arithmetic.
The navigation algorithm currently is implemented using key frames as a means for localization.
I also built the servo motors electronics and implemented the servo motors embedded
control software, as well as designed, 3D printed and put together the full body of the latest prototype.
Recommendations received
4 people have recommended Roumen
Join now to viewMore activity by Roumen
-
Gary Marcus is absolutely worth paying attention to, not because he’s controversial (he isn’t, imo), but because he shines a light on the…
Gary Marcus is absolutely worth paying attention to, not because he’s controversial (he isn’t, imo), but because he shines a light on the…
Liked by Roumen Popov
People also viewed
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More