Semantic Image Segmentation based on Convolutional Neural Networks (CNNs) is one of the main approaches in computer vision area. In convolutional neural network-based approaches, a pre-trained CNN which is trained on the large Image classification datasets is generally used as a backend to extract features (Image descriptors) from the Images. Whereas, the special size of output features from CNN backends are smaller than the input Images, by stacking multiple deconvolutional layers to the last layer of backend network, the dimension of output will be the same as the input Image. Segmentation using local Image descriptors without involving relationships between these local descriptors yield weak and uneven Segmentation results. Inspired by these observations, in this research we propose Context-Aware Features (CAF) unit. CAF unit generate Image-level features using local-Image descriptors. This unit can be integrated into different Semantic Image Segmentation architectures. In this study, by adding the proposed CAF unit to the Fully Convolutional Network (FCN) and DeepLab-v3-plus base architectures, the FCN-CAF and DeepLab-v3-plus-CAF architectures are proposed respectively. PASCAL VOC2012 datasets have been used to train the proposed architectures. Experimental results show that the proposed architectures have 2. 7% and 1. 81% accuracy improvement (mIoU) compared to the related basic architectures, respectively.

Recent researches on pixel-wise Semantic Segmentation use deep neural networks to improve accuracy and speed of these networks in order to increase the efficiency in practical applications such as automatic driving. These approaches have used deep architecture to predict pixel tags, but the obtained results seem to be undesirable. The reason for these unacceptable results is mainly due to the existence of max pooling operators, which reduces the resolution of the feature maps. In this paper, we present a convolutional neural network composed of encoder-decoder segments based on successful SegNet network. The encoder section has a depth of 2, which in the first part has 5 convolutional layers, in which each layer has 64 filters with dimensions of 3×3. In the decoding section, the dimensions of the decoding filters are adjusted according to the convolutions used at each step of the encoding. So, at each step, 64 filters with the size of 3×3 are used for coding where the weights of these filters are adjusted by network training and adapted to the educational data. Due to having the low depth of 2, and the low number of parameters in proposed network, the speed and the accuracy improve compared to the popular networks such as SegNet and DeepLab. For the CamVid dataset, after a total of 60, 000 iterations, we obtain the 91% for global accuracy, which indicates improvements in the efficiency of proposed method.

Image Segmentation is a fundamental approach in the field of Image processing and based on user’ sapplication. This paper propose an original and simple Segmentation strategy based on the EM approach thatresolves many informatics problems about hyperspectral Images which are observed by airborne sensors. In afirst step, to simplify the input color textured Image into a color Image without texture. The final Segmentationis simply achieved by a spatially color Segmentation using feature vector with the set of color valuescontained around the pixel to be classified with some mathematical equations. The spatial constraint allowstaking into account the inherent spatial relationships of any Image and its color. This approach provideseffective PSNR for the segmented Image. These results have the better performance as the segmented Imagesare compared with Watershed & Region Growing Algorithm and provide effective Segmentation for theSpectral Images & Medical Images.

In this paper the method used for Image compression is based on successive Image Segmentation. The Image to be compressed is segmented into smaller regions and this process is continued until each region can be approximated by a desired parameter so that the error be small enough in each region. In general, each region is an n-gon n where 3 < n < 8 . This method has short coding and decoding times. The coder is tree-structured, thus it is suitable for progressive Image coding.

Stochastic models such as mixture models, graphical models, Markov random fields and hidden Markov models have key role in probabilistic data analysis. In this paper, we used Gaussian mixture model to the pixels of an Image. The parameters of the model were estimated by EM-algorithm.In addition pixel labeling corresponded to each pixel of true Image was made by Bayes rule. In fact, a new numerically method was introduced for finding the maximum a posterior estimation by using EM-algorithm and Gaussians mixture distribution. In this algorithm, we were made a sequence of priors; posteriors were made and then converged to a posterior probability that is called the reference posterior probability. Maximum a posterior estimated can determine by the reference posterior probability which can make labeled Image. This labeled Image shows our segmented Image with reduced noises. We presented this method in several experiments.