The medical image compression plays an important role in digital image compression. Compression reduces the redundancy and encodes the data. To use different techniques and algorithms we can get the desired result. But it will depend on the used algorithm, which algorithm we have used in which circumstances. In this paper we have used SPIHT method with wavelet bior4.4 on brain surface medical image with different type of image parameters. And observe the result of this method with different types of image parameters at different level of iterations.

The MRI brain image detection and classification is very crucial for medical science. From last few years, various methods are being presented. Here in this paper, we proposed a novel approach to classify a given MRI brain image as normal or abnormal i.e. affected through tumor. The performed method firstly taken the wavelet transform for feature extraction from the images then applied principle component analysis to decrease the dimensions of features. Then the decreased features are being employed to kernel-support vector machine (k-SVM). Here the data has been taken for seven general brain diseases viz. glioma, Alzheimer's disease, Sarcoma, coma, LINN, IPOL, HPOL. We collected the data of 20 normal and 140 abnormal patients from Harvard Medical Laboratory website. Here four methods of SVM are performed such as L-Kernel, RBF, Poly, and NN. The comparative analysis is also being performed by the existing literatures which illustrates the result by DWT+PCA+KSVM by applying with GRB Kernel. Our classification gives fast and accurate results and assist doctors to distinguish between normal or abnormal to certain precision values.

Face alignment has been neatly done using deep convolutional neural networks. For achieving desired performance, the selecting strategy of local patches for training and their input range is crucial along with configuration of each network. There is double benefit of this. First, the texture background data over the full face is exploited to detect every key point. Second, the training has been given to networks to forecast every key point together; the geometric constraints amid key points are essentially encoded. So, the method can avert local minimum created by uncertainty and data corruption in challenging image samples because of occlusions, broad pose deviations and extreme lightings. Experimental results prove that the approach has high accuracy and is more robust and less complex in nature.