|
|
|
MHA IVRS is a subsystem of Ministry of Home Affairs. Entities of this subsystem include Conversant Integrated Voice Response system and a Windows NT server termed Simple Management Terminal (SMT). The functions of Conversant IVR include performing calls to various equipments like numeric pagers, alphanumeric pagers, and mobile phones, gather response, and voice administration. PL/SQL, C Programming, and shell programming are skills used to supplement the proprietary development tool. IHPT emphasizes on practical aspects of analysis, design and construction, testing, reliability and documentation. The work undertaken will further reinforces theories and concepts. The four phases are: Phase 1: Hardware
Phase 2: Application Phase 3: Design Software Phase 4 (Project) Back to top. FYP was carried through a span of 9 months. The title of the project is “Comparison of Neural Approaches to Image Compression”. The project was undertaken without any prior knowledge on image compression and neural networks. The project was carried out in 5 distinct phases, each implementing one neural algorithm. The 5 algorithms implemented are Kohonen Vector Quantization, K-Mean Vector Quantization, K-Means Scalar Quantization, Principal Component Analysis (PCA), and Independent Component Analysis (ICA). A standard lena image in PGM format is used as input for the various algorithms. The image is as shown below:
During each phase, signal-to-noise ratio and compression ratio were used as metrics for comparison. The neural algorithms were used to train a chosen neural network architecture so that the network would adopt the properties of the given inputs. Vector quantization is main approach used to achieve image compression. Back to top. |