Tetyana Baydyk

Recognition of pests on crops with a random subspace classifier

The purpose of this study is to develop and test a recognition system for the Colorado potato beetle. This task is very important for localizing the beetles and reducing the pesticide volume used to protect the harvest. We employ a beetle image dataset that contains 25 images representing different beetle positions and varying numbers of beetles. These images were collected from the Internet. Our recognition system is based on a special neural network, the random subspace classifier (RSC). We calculate the brightness, contrast, and contour orientation histograms of the images and use them as features and inputs to the RSC neural classifier. In addition, we describe the RSC structure and algorithms and analyse the obtained results. We obtained the best recognition rate of 85%.

Face recognition using special neural networks

The permutation coding neural classifier (PCNC) and its application to face recognition are described. Face recognition is a very important task in security applications such as those in airports and other institutions and for authentication via photo tagging in social networks. It is possible to improve face recognition with the PCNC using special types of distortion applied to the original images (for example, shifting of the images) in the training of the neural network. These distortions are applied to images in an initial image database (the FEI image database). The experiments are performed to test the performance of the PCNC on the database including the distorted image versions. We describe the FEI image database and report the image recognition results obtained.

Research and Development of a Recognition System for Measurements in Micromechanics

The article includes a description of research results in the areas of neural networks and image recognition with applications in micro mechanics. The principle objective is to develop a recognition system of edges of micro components for use in measuring sizes of these micro components. In applications of this recognition system in real control systems the elapsed time is a critical point. Therefore, one of the lines of research is to implement the neural network in electronic schemes. To implement algorithms of neural networks we used the electronic schemes ALTERA.

Computer Vision Based Manufacture of Solar Concentrator

Recently we developed flat facet solar concentrators that have low material cost. Each concentrator contains hundreds of components. To obtain low labor cost it is necessary to develop automated methods of components manufacture and assembly. In this paper we consider the problems of the concentrator manufacture. To make the automation algorithms more efficient we use computer vision methods based on neural networks. We developed these methods for handwritten digits and face recognition. Here we apply them to automated manufacture of solar concentrators. Several years ago we developed an algorithm of automatic placement of a pin in a hole. Later a visual based algorithm of component measurement was developed. In this paper we consider the possibilities of application of computer vision algorithms in solar concentrator manufacture.