By developing connections and increasing the use of the internet in today’s world, the necessity of secure connections as well as the assurance of inaccessibility of precious data is increasingly important. In this respect, studies on the science of data hiding have been widely done. In general, data security systems are classified into encryption and data hiding categories. One of the main branches of data hiding is steganography which its main purpose is hiding the existence of data from enemy or other users. Steganography is used in different media such as images, texts, videos and voices. However, improving steganalysis methods is essential for evaluating and applicability of steganography methods and preventing illegal uses. To achieve this aim, first, the previous methods of steganography and steganalysis should be studied. Then, given the changes that are made in the stego, steganalysis algorithms are expressed. In the present study, general explanations about the steganalysis algorithms in different areas are presented. In addition, it is tried to provide general information about the algorithms. Furthermore, according to the previous studies, it is attempted to present an improvement to their result’s methods. In order to make feature set, discrete wavelet transform is used for considering all images. Since, the dimensions of feature set increase and the classification is done in a long time, in order to decrease the dimension of feature set, LDA algorithm is used in the study. Results showed that this algorithm had a great impact on the feature set. That is, it decreased the dimension and increased the precision of recognition. Keywords: Steganography, Steganalysis, Feature set, Wavelet transform