No abstract

There has been a focus on developing image indexing techniques which have the capability to retrieve image based on their contents. The technologies are now generally referred to as Content-Based Image Retrieval (CBIR). Due to its extensive potential applications, CBIR has attracted a great amount of attention in recent years. Using colors as the content, content based image processing have been carried out for a sample of high resolution urban image and low resolution rural image scenes obtained from satellites. The color based processing has been utilized to identity important urban features such as buildings and gardens and rural features such as natural vegetation, water bodies and fields applying various techniques. The techniques included color based extractions using neighborhood rules and histograms. An estimation of the features and available resources from the imageries have been made using the color spectral graphs. The results of the analysis are presented and discussed in the paper.

The properties of Wavelet Transform can be successfully applied for analysis and processing of non stationary signals e.g., speech and image processing, data compression and communications. Due to the growing number of applications in various areas, it is necessary to explore the hardware implementation options of the Discrete Wavelet Transform (DWT). The Wavelet Series is just a sampled version of Continuous Wavelet Transform (CWT) and its computation may consume significant amount of time and resources, depending on the required resolution. The Discrete Wavelet Transform (DWT), based on sub-band coding, is a fast computation technique of Wavelet Transform, easy to implement and reduces the computational time and resources. Wavelet Transform uses multi-resolution technique by which different frequencies are analyzed with different resolutions. The study of 2-D DWT architectures reveals that there are two schemes for implementing DWT, one is based on convolution and other is based on lifting scheme. In this paper detailed study of Lifting Scheme has been carried out. Different architectures have been studied and performance parameters such as PSNR and Compression Ratio are determined. After obtaining double precision value of the image of size 256*256 imagery in BMP format, discrete wavelet transforms techniques are applied to obtain the wavelet coefficients for calculating PSNR and Compression ratio. Inverse Discrete wavelet transform are applied to get back the reconstructed image. It is found that for both satellite Rural and Urban imageries, the lifting scheme is very useful for obtaining higher quality of reconstructed images while achieving better PSNR~29 and Compression Ratios ~8.

No abstract

Image compression methods employing wavelet transforms have been successfully implemented to provide high compression rates while maintaining good image quality. The main contribution of wavelet theory and multi-resolution analysis is that, it provides an elegant framework in which both anomalies (such as edges and object boundaries) and trends (areas of high statistical spatial correlation) can be analyzed on an equal footing This results in a considerable improvement in encoding the significance map, and hence, a higher efficiency in compression. By employing the Successive approximation entropy coded quantization (SAQ), the EZW (Embedded Zero-tree Wavelet) coder generates a representation of the image that is coarser-to-finer in both the spatial domain and frequency domain simultaneously. Applying the DWT coefficients and EZW techniques using five threshold values, the maximum compression ratios achieved for an acceptable quality of the image have derived. The three types of images LENA Image and high resolution urban image (SatUImg and low resolution Rural (SatRImg) image have been considered for the analysis. The results show that for the standard Lena image one can achieve highest compression ratios (~11.28). where as the Rural Images (SatRImg) shows better compression ratio (~5.75) compared to that of Urban Images (satUImg) which is small (~1.39). The results are presented and discussed in the paper.