Identification and classification of plant species is a very important process in plant science and is widely used in botanical, medicinal and agricultural sciences. Traditional identification of plants is based on observations of their morphological characteristics of leaves, fruits, flowers, stems and roots. Molecular methods are also one of the modern methods that are more accurate than traditional ones, but they are very costly. The leaf is an important part of the plant, which available in most seasons. Accordingly identification and classification of plants may be done based on leaf features. One of the most important agricultural products in the world is potato with nearly five thousand different species, each of them suitable for a special kind of weather or soil conditions, as well as for various uses. One of the important issues involved in the preparation and production of seed potatoes of different classes is the problem of mixing different species of potato tubers. This may lead to considerable economic losses. Image processing and pattern recognition methodes, may be used in automated identification and classification of plant species. One of the requirements of agricultural industry is to classify ssed potatoes. In this thesis, using image processing methods, an efficient algorithm for identifying and classifying eight different types of potatoes is presented using seedlings. The database consists of 373 images of seedlings of eight different types of potatoes that have been produced with the collaboration of the Agricultural Biotechnology Research Institute of the Central Region, Isfahan, Iran, and several other greenhouse producing SE seed potato tubers in the region. The features extracted from the leaf images include shape, texture and color of the leaves. Each of the extracted features is evaluated individually and in different combinations. Various potato varieties in the database is determined by using a combination of features. For this the results of different classifiers are fused based on probability matrix. The method succeeds in classifying the species in a database with accuracy of 96.79%. Keywords: Leaf recognition, Plant classification, Feature extraction, Seedling image processing