Scientific Research

Krendls Ltd develops complex applications suggesting innovation solutions. Coping with task of any level is a strong point of the Krendls scientific research department, represented by world-known PhD and Full Professor scientists.

The principle exploration directions of the scientific group:

• Multivariate signal processing and compression
• E-documents processing and compression
• Video sequences colorization
• Mono to stereo image conversion
• Signal quality enhancement
• Geoinformation system data processing
• Numeric program unit machines software development
• Cam mechanism analysis and synthesis problem solution

Multivariate signal processing and compression
The most informative signal part separation depending on its origin. For example, photo and video processing and compression problems, medical images (roentgenograms, computer tomograms), multi-zone satellite images handling, spectral qualities enhancement.
The group has developed a number of methods, algorithms and software applications with various readiness degree for processing such signals as:

• Static images
• Video sequences
• Medical images
• Multi-zone satellite images

In particular, we have developed methods of data processing and compression algorithms, oriented to various application domains, namely:

• fast encoding methods (for aerospace filming data processing usages)
• fast decoding (for remote client computers usage and video sequence reproduction)
• adaptive encoding and decoding techniques (in problems that require regarding of processed data peculiarities)
• Lossless encoding methods (in the problems where data loss is unacceptable, for instance, navigational, medical data, etc.)

Large data amounts encoding methods (block encoding problems and tomographic, aerospace video and other data encoding and reproduction in different planes). The acquired wavelet filters unlike the traditional ones, consider the specificity on the processed signal and compensate errors emerged on the previous processing stages. Besides the wavelet analysis we use signal separation into principle components with linear and non-linear methods.

E-documents processing and compression
Electronic document compression, page data transfer, unauthorized copy information protection (copyright protection).
The group has developed a software package for electronic documents processing and compression. The basis of the package is:

• Electronic document separation into informative layers (symbol layer, tables, graphs, raster layer and background)
• A corresponding compressor application to every layer
• Block (segment) data representation, reproducing documents partially (like separate layers as well as pages)

Software package features:

• Block document transfer via the Internet or a LAN and its reproduction via the Internet Explorer
• Impossibility of saving a document as a whole and its autonomous usage (unauthorized copy protection and copyright security)

Local package version gives possibility to assemble several documents into one collating with navigation, margin marks and so forth.

Video sequences colorization
Black and white movies revival and colorization problem (and photography as a special case).
We have developed a software package for colorization and revival of black and white movies. The basis of the package is:

• Automated movie separation into scenes
• Flexible and configured system hierarchy of colour templates
• Developed methods of two- and three-dimensional colorization according to user markings
• User logics prediction (user markings)

Package features:

• User permission separation allows several users to work correctly on a project (movie)
• Interactive modifications of colorization process (colour correction, lighting, template replacement, etc.).

The package developed permits to implement a fast and effective colorization of films by any number of employees.

Mono to stereo image conversion
Pseudo-stereo photos creation problem and movies from the flat originals.
The following methods are built:

• Object selection by user markings
• Image depth map and separate object formation methods
• Topological conversions of objects by the depth map
• Pseudo-stereo image construction from the two-dimensional analysis by the depth map

Signal Quality Enhancement
Signal filtering from regular and irregular noise. For example, video sequence quality improvement on the film with mechanical and chemical damage, document quality improvement (noise filtration).
We have developed a number of signal cleaning methods, based upon high-frequency signal component analysis. For electronic signals noise cleaning there have been developed proper methods and algorithms considering local peculiarities of regular signal.

Geoinformation system data processing
Contour and surfaces building and description using minimal number of informational characteristics, adaptive triangular grate building.
There have been developed such methods as:

• Contour and surface construction and description using minimal number of information characteristics
• Construction of adaptive triangular grates
• Separation into information levels
• Compression and reproduction

Numeric program unit machines software development
Contour and trajectories construction from the technological curves with minimal frame number. Such curves can be splines (and polylines special case), circle arcs, logarithmical spirals arcs and circle scans, etc.
We developed methods of optimization of curve description with splines and technological curves with smoothly welded circle arcs (which is crucial for high-quality surface processing), smooth parts of circle evolvent (airplanes, automobiles and ships movement trajectories), logarithmic spirals (frames), etc. A certain share of the results is embedded into "Electronmash" production union (St. Petersburg, Russia).

Cam mechanism analysis and synthesis problem solution
Existing cam mechanisms synthesis and analysis and cam disk profile analytical construction problem solution according to the given pusher movement law. It's possible to describe a disk with a piecewise circle curves with the minimal frame number. Contour and trajectory construction from the technological curves with minimal frame number.
On the analytical description basis according to the required law of a pusher movement an analytical description of cam disk profile is constructed. The problem is solved for different kinds of pushers (flat, needle, frame, ect.)

Multivariate signal processing and compression
The group has developed a number of methods, algorithms and software applications with various readiness degree for processing such signals as:

• Static images
• Video sequences
• Medical images
• Multi-zone satellite images

In particular, we have developed methods of data processing and compression algorithms, oriented to various application domains, namely:

• fast encoding methods (for aerospace filming data processing usages)
  
• fast decoding (for remote client computers usage and video sequence reproduction)
  
• adaptive encoding and decoding techniques (in problems that require regarding of processed data peculiarities)
  
• Lossless encoding methods (in the problems where data loss is unacceptable, for instance, navigational, medical data, etc.)

Large data amounts encoding methods (block encoding problems and tomographic, aerospace video and other data encoding and reproduction in different planes). The acquired wavelet filters unlike the traditional ones, consider the specificity on the processed signal and compensate errors emerged on the previous processing stages. Besides the wavelet analysis we use signal separation into principle components with linear and non-linear methods.