Research On Signal Processing Method For The Analysis Of Wheat Quality

Grain security is one of the major factors that are related to China's social stability and economic development as well as its national security. In a country that feeds 22 percent of the world's population with only 7 percent of the world's arable land, grain security stands at the top of the government's agenda. Grain security requires both quantity and quality. Grain reserves of a proper scale, structure and high quality are needed, which is a necessary measure to protect national grain security. Wheat is one of the most important food grains in China. To enhance quality of wheat, reduce postharvest loss and ensure grain security, great efforts should be made to develop rapid, objective, and easy methods for assessing wheat quality using signal processing techniques.Various quality parameters, including end-use quality, may be obtained by using corresponding standard physical and/or chemical methods. However, it is usually difficult for inspectors to use these methods for assessing wheat quality in routine inspection because procedures may be time-consuming and often require various expensive instruments. While those who are in charge of the inspection may be familiar with the quality of all varieties of wheat involved, one alternative to the conventional methods is to recognize wheat sample varieties.Color is one of the most important characteristics of wheat varieties among all of the kernel features that are available for evaluating wheat varieties by visual assessment. The color feature could be measured conveniently and is invariant to the transformation in image that is due to rescaling, translation, distortion, and rotation. Based on the analysis of researches on color features in the visible, near-infrared (NIR) reflectance spectra methods, it could be concluded that it is not unexpected that color features used in some researches are too imprecise to produce satisfactory recognition rates for varietal recognition. Color histogram is one of the most widely used tools in image processing, image retrieval, computer vision and so on. As the classical one-dimensional indexing structures are not appropriate to three-dimensional color indexing, the generation of color histogram is one of the basic problems in these researches. A fast and efficient index structure for generation of color histogram is highly desirable andbeneficial. A novel index structure, the Sparse Forest (SF), for generation of color histogram is proposed based on spatial data indexing. By projecting colors on the (r,g,0) plane, theproblem of three-dimensional color indexing is reduced to one-dimensional indexing, which is performed using B-trees. This index structure is capable of preserving color spatial information and tends to provide excellent computational performance and good space utilization. The analytical and experimental results are presented, showing that the index structure compares favorably with the traditional spatial data structures in terms of overall algorithm complexity in the case of color histogram. This paper develops a rapid, objective, and easy method for recognizing wheat varieties, which is important for breeding, milling and marketing. The method can be used in place of the existing procedures to remove subjectivity from wheat variety recognition. In contrast to previous work, most of which has focused on wheat morphological characteristics, the features utilized in this paper are based on kernel color. Varietal classification is performed by using Support Vector Machines (SVM) method. More than 96% correct recognition rates are achieved with bulk samples involving 16 varieties representing a wide range of wheat varieties, wheat class, and kernel types. The proportion of single wheat kernels correctly recognized ranges from 87% to 93%. The results were encouraging since the method proposed here can be easily conducted in routine inspection.Reducing additive noise by using independent component analysis from one noisy signal can be rephrased as the application of ICA under insufficient observations. The key to the problem is to construct another mixed signal reasonably. Periodic noises represent the principal disturbances in the form of an additive noise independent of the wheat quality signals. In this paper, a novel method for constructing a mixed signal is proposed based on the holistic periodicity of periodic noises. Much better signal-to-noise ratio (SNR) could be obtained by using ICA with the constructed mixed signal. The algorithm presented in this paper proves to be fast, computationally simple and efficient. Simulation and experimental results demonstrate the feasibility and good performance of the proposed algorithm.Blind deconvolution (BD) is a fundamental problem in image processing, speech signal processing, communication, system identification and so on, which has attracted extensive research interests in the past decade. In a traditional homomorphic deconvolution system, linear filtering is often used. This paper presents a novel blind deconvolution algorithm, in whichindependent component analysis (ICA) instead of linear filtering is applied to separate the complex cepstrum signals. The algorithm provided is fast and capable of separating the signals with high precision. Simulation and experimental results demonstrate the method for blind deconvolution is effective.The impact force method is used to acquire wheat quality signal, which is buried by periodic noise and disturbances originated from the environment and from the measurement system itself. Noise reduction is first performed by the algorithm for periodic noise reduction based on ICA, followed by deconvolution with the blind deconvolution algorithm presented in the paper. The features Z (average cross zero ratio), I (average signal length) and SKQ (crosszero number) are introduced to serve as sample features in analysis of the bulk and single kernel wheat quality respectively. Compared to other analysis methods, though less sample is required, the impact force method classify all the bulk wheat samples correctly, and is capable of evaluating single kernel wheat quality. The impact force method for the analysis of wheat quality has being applied for China invention patent, which is in the patent examining procedure now.