An Optimized Method for B-Mode Echocardiographic Video Compression Based on Motion Estimation and Wavelet

In this paper, a new approach for echocardiography image compression is developed. To achieve a high rate of image compression as well as preserving the image information, motion detection and wavelet transform are combined. In the first step, a Region Of Interest (ROI) is determined and the image is divided into several (8
8 pixels) blocks. Thereafter, the motion vectors of each block are estimated to predict the subsequent frame (predicted model frame). Additionally, the wavelet component is created by applying the wavelet transform to the main image (which should be predicted); whereas the extracted wavelet component is utilized as a predicted frame error compensator. Subsequently, entropy of the motion vectors of each block is extracted as a criterion to determine the level of quantization which is used for wavelet frame quantization. Wavelet frame is quantized based on the Lloyd's algorithm. Finally, the correlation of each block of the predicted model frame with the corresponding block of the main image is calculated to evaluate the rate of the accuracy of the result. If the calculated correlation is more than 0.5, an optimized combination of the predicted model and the corresponding block of the wavelet frame is utilized as the final block. Otherwise, the block of the wavelet frame is considered as the final result. The results were analyzed using PSNR, MSE, GLCM and expert-based quality image validation. The proposed algorithm was compared to standard MPEG standards, H.264 and VC1 which proved the out-performance of the proposed algorithm.