Research And Implementation Of The Algorithm For Dental Colorimetric Based On DSP

In recent years, with the color of porcelain teeth are much more similar to the natural teeth, they are more and more concerned by people. Clinically, the selection of the porcelain tooth color mainly relies on the color model through dentists’ visual perception selecting the most match color between the shade guide and the tooth color of patient. The color selected by dentist has big impact due to individual differences, so the researchers committed to the use of machine vision technology to realize the computer to help dentists choose color.Based on the previous probing and colorimetric algorithms, this paper designs a tooth color matching and analysis system based on DSP. The system uses ICETEK-DM6437-B-KIT of Realtime Technology Co. as the hardware platform. This system includes teeth image preprocessing, feature extraction, classification and identification of the function and it realizes the template images display after identifying categories. The main achievements can be described as follows.(1) Building up dental sample library. This paper uses the image acquisition equipment to get teeth images and builds up dental sample library based on the wide used clinical application of Vita3D Master dental shade guide.(2) Methods for teeth images preprocessing. In this paper, teeth are separated from the background according to the unique features of the captured teeth images. As there are some reflection areas on the surface of a tooth, forward squares weighted methods was used to get the initial data according to different importance of each part of the tooth to the tooth colorimetric.(3) Feature extraction and dimensionality reduction of the initial data. Based on the previous research of colorimetric, this paper has adopted Principal Component Analysis (PCA) and Two Dimensional Principal Component Analysis (2DPCA) to achieve feature extraction, and acquire a feature projection space. The two methods make a significant reduction of the calculation cost of the DSP system, and ensure the accuracy of the classification.(4) Achieved classification of porcelain teeth. According to the MATLAB simulation experiment results of the two kinds of classifications-K-nearest neighbor and support vector machine, this paper choose K-nearest neighbor classifier finally.(5) Achieved the basic function of the system on DSP hardware. By using C language, achieved the dental image classification in CCS, and got the success of writing test in DSP application board.Experimental results show that the algorithms designed in this thesis robust, and this algorithm provides technical basis for dental colorimetric system.In the end, the whole work of this paper is summarized and the future work is planned.