Study On Selection Method Of The Pipa Front Board Based On Texture Images

At present,the selection of wood used in the soundboard production industry in China's clam production industry mainly relies on the musical instrument technicians' "view","deaf" and "hear"methods to subjectively judge.Among them,"view" refers to the production technician to visually observe the straightness of the texture of the board to be selected,the pitch of the texture,and the like to determine the merits of the board.However,this "view" selection method has a relatively high requirement on the personal experience of the technician.This traditional selection method lacks the guidance of scientific theory and is not conducive to the inheritance of the production process.At the same time,the traditional method of material selection restricts the improvement of vocal quality of pipa,the accuracy of material selection and the rate of material production.According to the?Sawn timber for national music instrument-Part 2:Timber for pipa?,promulgated by the State Forestry Administration in 2011,and combined with previous research results,it can be found that in the process of making pipa,we should choose the wooden planks without scarring,unmistakable cracks and moth-borers.In this paper,the preference of this traditional national musical instrument board is studied.According to the requirements of the industry production standards,the texture range of the front board should be between 1.8 and 2.5cm.At the same time,the texture of the board is required to be straight,and the inclination of the texture does not exceed 5 degrees.According to the judging standard of the pipa board,the board with less than 1%texture inclination can be used to make the first grade musical instrument,and the board with less than 3%texture inclination can be used to make the second grade musical instrument.Take the Tianjin First National Instrument Factory as an example,during the production process of making pipa,the technician determines whether the board is qualified and the grade it can be used to make the instrument by "view" the board texture.Through research,it has been found that although the musical instrument production technicians have a wealth of experience in material selection,but the texture of the board changes more complex,the naked eye observation of the plank will produce a larger error,affecting the selection of the judge.(1)Aimed to the defects in the current traditional "view" texture selection method,this paper first uses the edge detection method to perform edge extraction on the image of the panel material to enlarge the texture features of the image,so that the production technician can convert from the"view" plate texture to the "view" text image after magnifying the texture features,which is helper for technician to determine the straightness and other features of board texture more clearer.Experiments show that this method can improve the success rate of front board selection.(2)Concerning the problem of large spacing measurement error caused by the density of the board and the fixed spacing,this paper proposes a gray-scale projection method to quantify the texture of the board by plotting the gray projection curve of the front board material.And using the measuring ruler to obtain the texture of the front board spacing values and average spacing values.Comparing the texture pitch value obtained by using this method with the industry standard,and you can identify whether the board is qualified or not.(3)Because it is necessary to use an edge detection algorithm to perform edge extraction on the board image,this paper analyzes the traditional edge detection operators such as Sobel,Laplace,Canny and other methods that have been improved in recent years.It is found that the traditional Canny edge detection algorithm has the disadvantages of denoising and preserving the edges,and the improved algorithm of other people can not meet the needs of this paper in terms of accuracy.Therefore,the paper improves the edge detection operator and uses the wavelet threshold function based on the generalized cross validation criterion to replace the Gaussian kernel image in the Canny operator for smoothing and denoising.For the problem that the traditional Canny operator needs to be set manually,the Otsu algorithm is used to implement adaptive threshold acquisition.For the performance of the improved algorithm,the evaluation criteria of edge detection is introduced for evaluation.This paper compares the results of artificial "view" textures,and compares the accuracy of the optimized Canny operator before and after improvement.