Use Of Discrete Cosine Transform Research Articles

Quality Improvement in Color Image Compression using New FDCT and FIDCT

image compression methods aim to compress color image while ensuring that noise removal compression technique produce the good quality level images. The storage of images is becoming difficult with number of images growing to million and billions. Hence the image compression becoming absolute necessity in computing field. The Discrete Cosine Transform is widely used in image compression technique. Redundant information in an image needs to be eliminated by adopting intelligent method. In the recent research attempts better quality of compression is observed with the use of Discrete Cosine Transform (DCT) and Quantization. In this paper, the proposed system efficient image encoder removes the sinusoidal noise from original image by Butterworth Band Reject Filter (BBRF) algorithm in FDCT color image compression. The objective of this color image compression scheme is efficiently noise removal image, calculate the Compression Ratio(CR), Peak Signal to Noise Ratio (PSNR) Mean Square Error (MSE) by changing the FDCT level and using Fine Inverse Discrete Cosine Transform (F-IDCT) factor while preserving the quality of reconstructed image. Experimentation has been carried out on different image formats successfully. The proposed system achieved a good compression ratio and considerable test application for the quality of the reconstructed color image. KeywordsFIDCT, PSNR, MSE, Sinusoidal Noise, BBRF, Image Compression.

Analysis of Single-Electrode EEG Rhythms Using MATLAB to Elicit Correlation with Cognitive Stress

This paper demonstrates electroencephalogram (EEG) analysis in MATLAB environment with the objective to investigate effectiveness of cognitive stress recognition algorithm using EEG from single-electrode BCI. 25 subjects' EEG were recorded in MATLAB with the use of Stroop color-word test as stress inducer. Questionnaire on subjects' self-perceived stress scale during Stroop test were gathered as classification's target output. The main analysis tool used were MATLAB, coupled with the use of Discrete Cosine Transform (DCT) as dimension reduction technique to reduce data size down to 2% of the origin. Three pattern classification algorithms' - Artificial Neural Network (ANN), k-Nearest Neighbor (KNN) and Linear Discriminant Analysis (LDA) were trained using the resulted 2% DCT coefficients. Our study discovered the use of DCT along with KNN offers highest average classification rate of 72% compared to ANN and LDA.

Design, analysis and experimental evaluation of block based transformation in MFCC computation for speaker recognition

Standard Mel frequency cepstrum coefficient (MFCC) computation technique utilizes discrete cosine transform (DCT) for decorrelating log energies of filter bank output. The use of DCT is reasonable here as the covariance matrix of Mel filter bank log energy (MFLE) can be compared with that of highly correlated Markov-I process. This full-band based MFCC computation technique where each of the filter bank output has contribution to all coefficients, has two main disadvantages. First, the covariance matrix of the log energies does not exactly follow Markov-I property. Second, full-band based MFCC feature gets severely degraded when speech signal is corrupted with narrow-band channel noise, though few filter bank outputs may remain unaffected. In this work, we have studied a class of linear transformation techniques based on block wise transformation of MFLE which effectively decorrelate the filter bank log energies and also capture speech information in an efficient manner. A thorough study has been carried out on the block based transformation approach by investigating a new partitioning technique that highlights associated advantages. This article also reports a novel feature extraction scheme which captures complementary information to wide band information; that otherwise remains undetected by standard MFCC and proposed block transform (BT) techniques. The proposed features are evaluated on NIST SRE databases using Gaussian mixture model-universal background model (GMM-UBM) based speaker recognition system. We have obtained significant performance improvement over baseline features for both matched and mismatched condition, also for standard and narrow-band noises. The proposed method achieves significant performance improvement in presence of narrow-band noise when clubbed with missing feature theory based score computation scheme.