Dissertation > Excellent graduate degree dissertation topics show

Color Constancy Algorithm Based on BP Neural Network

Author: JiangLong
Tutor: Cai
School: Shandong University
Course: Applied Computer Technology
Keywords: Color constancy Neural Networks Machine Vision Background subtraction
CLC: TP391.41
Type: Master's thesis
Year: 2010
Downloads: 116
Quote: 2
Read: Download Dissertation

Abstract


With the continuous development of information science and technology, digital image technology has made great progress in many fields of military, agriculture, transportation, aerospace, sports and shows great potential for development. The color image processing technology, has become a hot research topic in the field of machine vision content. Target recognition technology in the field of machine vision also video conferencing, security surveillance, content-based image and video retrieval field, the more widely used. Among these applications is a very important issue is to solve the problem of color recognition. For machine vision systems, the color is an important feature of the part of the object, and plays an important role in image understanding and object recognition, machine vision systems often need a stable surface spectral reflectance characteristics described in the application of color information, which relating to the problem of color constancy. Color constancy is not due to changes in the external environment and the ability to maintain the same object color perception. Generated by the light source gradients in machine vision, this paper presents a neural network-based color constancy algorithm. The first design a BP neural network, by optimizing the network parameters to improve the learning algorithm to overcome the traditional BP network is easy to fall into local minimum point, slow convergence defects. After extensive training of the neural network through appropriate training set, the mapping between the light changes before and after images of the corresponding pixels in order to establish the color constancy model in gradient light environment. This method does not require adaptive models are built-in constraints, the surface properties of the input data does not need to make specific assumptions have adaptive, self-learning characteristics. This article main innovation is to characterize the color of an object by the the Luv color model the characteristics of the human visual system color constancy, the color information of the original image obtained by machine vision systems converted based on the brightness of the image information and based on the color image information, the traditional color constancy problem is decomposed into brightness constancy and constancy of the color information. The article also on the the color constancy implementation mechanism made assumptions proposed color constancy by local result of the exchange of information by the visual system processor subsystem. To test the neural network model to eliminate the effect of light source gradient factors, natural light conditions, a large number of experimental verification of the model, experimental results show that the model of indoor real environment gradient daylight color recognition showed better color constancy, color constancy algorithm is a very effective method.

Related Dissertations

  1. High Speed Frequency Measurment and Non-Linearity Correction of Frequency Modulated Capacitive Displacement Sensor,TH822
  2. FPGA/DSP Image Co-processor Technology and Ethernet Data Transmission,TP391.41
  3. Research on Anti-periodic Solutions of Delayed Cellular Neural Networks without Assuming Global Lipschitz Conditions,TP183
  4. Designs and Applications of Fuzzy Synthetic Evaluation Models Based on Parallel Algorithms,TP18
  5. Research on the Theory of Visual Object Tracking Based on Particle Filter for Autonomous Robot,TP242
  6. Research on Detecting Optical Fiber Geometric Parameter Based on Machine Vision,TN253
  7. Spectrum Variation of Vegetation in Yanzhou Coal Mine Area and Heavy Metal Stress Characteristic,X173
  8. Intelligent video detection technology based on the traffic lights control,TM923.5
  9. Research into Visual Orientation in the Weeding Robot Based on COCM,TP391.41
  10. Multi- license plate location method based on CNN's Intelligent Transportation Systems Research,TP391.41
  11. Research on the Models of GPS Height Fitting Based on BP Neural Network and Their Applications,P228.4
  12. Research of QoS Optimization Based on Neural Network Prediction for Ethernet Passive Optical Network,TN929.1
  13. Sensitivity Study and Application of Complex Variabal Weight Function Neural Network,TP183
  14. Researches on the Application of the Neuron-MOS in the Neural Network,TP183
  15. A Reduction Method for Artificial Neural Network Inputs Based on An Improved Genetic Algorithm,TP18
  16. On-line Monitoring of Tool Wear for Mass Production,TG71
  17. Modeling of Photovoltaic Cells and Its Application in Energy Prediction,TM914.4
  18. The Design of Identification System for Power Marketing Services,TP391.41
  19. The Study of Intelligent Intrusion Detection System Based on Neural Network in Linux,TP393.08
  20. RFID Information Sharing in Supply Chain and Business Valuation Research,F224
  21. Consider short-term load forecasting PV,TM715

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile