The Research Of Intelligent Wood Species Recognization Based On DSP

Wood species is one of the important factors for the quality of the building board, especially the effect of it on the phisical mechanics performance. The main purpose of this research is to recognize the wood species quickly and accurately, so that the parameters in the next process can be controlled at real time and finally the quality of the building board can be improved.There were 108 images including 38 hardwoods' and 16 softwoods' for us to study how to recognise the wood species using the texture features and color features. First, we got the image from the CCD camera. Secondly, we got the digital image converted through A/D converter that could be pre-processed and analysed using compouter or DSP. Thirdly, we computed values of texture features and color features, including contrast, correlation, variance, energy, brightness ratio of Red, Green and Blue. At last, we finished the recognition by Back Propagation Network which was also compiled by C language and loaded the programme to the DSP chip. In this process we could finish the operations at real time on DSP.In this paper, we first introduced the research background, content and meaning. After comparing the image process systems on DSP nowadays, we put forward the hardware design method, in which design we used the mixed structure of FPGA and DSP that could improve not only accuracy of the image processing but also the processing performance. Next, we introduced the principal of the image acquisiton and some methods of the image processing, including image smoothing method and image enhancement method. We mainly introduced the theories of the texture features and color features extracted. We also introduced some image recognition methods, of which the ANN method was the important one. After the test, we put forward the BP network model, by which we got the recogniton result that the accuracy rate was above 96% and the steability was good. At last we analysed the reasons for the mistakes that the number of the training samples was not enough.The research is important for the wood production and utilization. If it is used in the production of flakeboard and other building boards, the quality can be improved, meanwhile, the wood can be taken good use of. In addtion, the hardware design put forward and the method of extracting features can be good reference for other image processing systems.