This thesis presents an overview of computer vision. It gives basic insight of computer representation of image and formation of the image in the camera. Brief history of OpenCV along with contents of the library are shown. It also covers basic functions and structures. Algorithms for removing noise and edge detection in the image are shown and explanation of morphological operations with their application example is given. There are examples of contours extraction and corner detection also. SIFT key point detection and extraction is explained along with performance comparison to SURF detector. The basic idea underlying classification of objects is shown along with explanation of cascade classifier algorithm. LBP and Haar features, which are available with OpenCV for cascade classifier detection, are presented. SVM is listed also. Practical work covers construction and implemented of algorithm for open palm detection and finger counting. Training data preparation for cascade classifier training with OpenCV tool is explained. Classifiers are trained on three sets with different number of positive examples and two different resolutions. For every set of positive images, classifier is trained for both LBP and HAAR feature and performance of resulting classifiers is compared. Some ideas for improving constructed algorithm and cascade classifier are given and last chapter is shown the possible acceleration via GPU units and additional OpenCV modules.