• R & D Services

    Image Processing

Digital Signal Processing and Image Processing

Uurmi provides design services in signal processing and Image Processing algorithm design and implementation of these innovative algorithms on Digital Signal Processors (TI, ADSP), and Micro Processors with DSP extensions (ARM, MIPS, FreeScale) using our excellent assembly language programming skills or can provide services with implementations in RTL hardware for FPGA’s or SoC’s. Uurmi differentiation comes from its ability to understand the platform configuration and characteristics, suitably modify or enhance the existing algorithms for a particular use case scenario, and optimize it for the target processor architecture. With vast experience in DSP architectures and instructions sets, customers can be ensured that the foot print and MCPS utilization is the best in class without losing performance.

Areas of expertise include the following:

  • Communication systems
    Physical layer algorithms for GSM/GPRS/EDGE/3G, WiFi, WiMax, especially DSSS, OFDM, MIMO Modulators/Demodulators like 16,64,256QAM, GMSK, QPSK Channel Coding – Viterbi, Convolution, LBCC, Reed Solomon, BCJ, CRC Synchronization – Tracking and Frequency Estimation, Tracking algorithms Channel Estimation – ML, LS/ZF, MMSE Adaptive array or smart antenna processing Adaptive Filters, Equalizers Security and Encryption Blocks

  • Audio, Video Codecs
    Uurmi has ported almost all the standard audio codecs used in VoIP – G.7xx codecs such as G.711, G.722 (wideband), G.723, G.729. We have also ported codecs suited to VoIP over public internet such as iLBC. The team has developed video codecs such as H.263, H.264, MPEG2 (decoders/encoders) in the past. We have the expertise to tune, modify, optimize and port any proprietary codec to an embedded microprocessor or DSP

  • Video Quality Monitoring
    Video Quality monitoring algorithms monitor the perceptual quality of the content to ensure a good end-user experience, understand where quality issues are injected and reduce support costs. It immensely help to contain costs by ensuring that customer complaints are handled quickly. It also ensure that operations and engineering are not incorrectly assuming that the problem is due to their processing or distribution of the channel in their own network.

  • Audio QoS/Enhancement
    Uurmi has the expertise in development of acoustic and line echo cancellation (AEC/LEC) algorithms, noise suppression and echo cancellation algorithms. We have ported these algorithms to ARM Processors and TI DSP and also optimized them for proprietary architectures.

  • Multimedia and Video effects
    Uurmi has the know-how on video effects such as Resizing, De-Interlacing, Color conversion, Gamma Correction, Brightness and Contrast control. We can customize and port the proprietary signal processing algorithms of our customers.

  • Image Processing and Computer Vision
    The teams expertise in defence and automotive sectors have given the exposure to develop the following algorithms for various applications

    Sample Technologies Sample Applications
    • Biologically inspired visual attention and gist
    • 2D and 3D object recognition
    • Evolutionary optimization methods
    • Texture analysis
    • Shape-based recognition
    • Morphology and Image enhancement operations
    • Uncertainty handling in images using fuzzy sets
    • Tracking and fingerprinting algorithms
    • Segmentation of intensity and range images
    • Uncertainty modeling in data sets
    • Context-aware target recognition
    • Machine learning and training of classifiers Sample Applications

      – Pedestrian Protection
      – Occupant sensing
      – Collision Avoidance
    • Vision-based safety in factories
    • Border/Force/Perimeter Protection,
    • Automatic Target Recognition
    • Training soldiers
    • Hidden weapon detection
    • Content-based retrieval
    • Radar image processing

    Sensors: Experience with EO/IR, mm-wave, hyperspectral, LIDAR

  • Cognitive Image Recognition System
    Uurmi is working on a video analytics project to capture, analyze and identify objects against a set of previously stored objects. The proposed algorithm is a collaboration between Uurmi and a customer. Furthermore, the system has the ability to identify unknown objects as a potential candidate for input to the recognition algorithm. Based on set of characteristics any such new identified object is stored in the input database.

Uurmi has an excellent team with many years of experience in porting and developing the signal processing and Image Processing software/firmware. Thus we are in a great position to take up turn-key projects of refactoring of existing DSP software modules and refactor them for optimal performance using hand coded assembly and any such processor specific features. Our talented team has great experience in developing software on TI, ADSP digital signal processor and proprietary configurable embedded processors and processors with multi-threading architecture.
Face Recognition System
Designed a complete face recognition and tracking system with more than 90% accuracy.
Case Study >>
Fill the form to download
Note: Document will be emailed to the specified email address.