Responsible for research, definition, design, implementation, and integration of the core algorithms in computer vision, including special and general object detection techniques.• 2D barcodes (DataMatrix, QRCode, Aztec, MaxiCode, DotCode, etc.) – multiple detection and decoding on still/video data with complex contexts. • Deep Learning for object detection and classification: training pre-trained YOLOv8 models for new object classes.• Integration with Android apps (.so lib, aar wrapper lib, client side).

• Mobile applications development (NOMO Lab, cloud based - AWS)• Machine Learning /Deep Learning • Python app development (scheduler)

Responsible for research, design and S/W simulation of core algorithms in front-end image processing chains to be implemented in H/W for digital camera systems as part of DSP. Row/column noise reduction, correlated double sampling, anti-shading, binning, Bayer downscaling, despeckle, wide dynamic range imaging etc. Interaction with Digital Design, Analog Design teams

Responsible for research, definition, design, implementation, and integration of the core algorithms in computer vision, image analysis and understanding. Estimating and improving quality of industrial machine vision oriented digital cameras.• 1D and 2D Barcode (Data Matrix) multiple(up to 300 objects per image) detection/decoding.Typical image characterization: Very small bar width (up to 1.1pixel) or matrix modules (up to 2.5x2.5 pixels) , non-uniform lighting and sharp shadows, high level of noise, projective distortion, insufficient focus at image periphery, complex background.Detection techniques: blob analysis, morphology, corner/line detection, oriented edge analysis, Hough transform, image pyramids, compensating for projective distortion.Decoding techniques: FFT, fuzzy models, classical and advanced techniques in decoding Reed-Solomon codes (beyond BCH boundary – Sudan’s algorithm, Koetter/Vardy approach).• Detecting/decoding color labels(small circular grouped labels divided into 10 colored segments).Blob analysis, segmentation/labeling, quad-tree decomposition, local white balance tuning and color normalization, ellipse fitting, segments localization, improving segments color sampling procedures, color aberration/aliasing compensation.

Responsible for definition, design, implementation, and integration of the essential algorithms for processing of still images (as captured by high-end digital cameras).Key player in infrastructure definition, architectural issues of the applications.Image Processor Engine module – core backend library, which includes different types of mosaic interpolation, image resampling (scaling), sharpening, transformation of raw image data in perceptual color space. Works in cross-platform (Macintosh and PC) environment. • Demosaicing library: interpolation of raw mosaic (Bayer pattern) image data, incorporating a set of adaptive filters, edge/line detectors, pattern matching, noise reduction, color smoothing and image sharpening. Multi-threaded, heavy duty.• Color Profile Builder (ICC) - a tool allowing editing input ICC profiles, ensuring precise color representation.• Interactive tools: gamut compression, non-linear gray balance, retouching brush, anti-moire brush (magicAL eraser).