Research On 3D Reconstruction And Evaluation Device Of Apple Quality

With the economic development and improvement of quality of life,consumers have a higher pursuit of apple quality but are unable to make effective judgments about whether the apples they purchase meet their taste needs.The current domestic apple quality evaluation methods have provided detailed industry evaluation standards to reduce the problem of uneven quality in the market.In terms of quality inspection,automated inspection instruments based on hyperspectral image technology have gradually replaced manual inspection to improve inspection efficiency,but the limited space for inspection and the high cost of most instruments make the inspection results somewhat limited.Therefore,it is of great value to design a low-cost3 D reconstruction device with hyperspectral detection function and the ability to combine multiple angles of apples for comprehensive quality evaluation.To address the above issues,this study contains the following main aspects.(1)Research on apple quality feature detection based on hyperspectral imaging.Comparing and analyzing the differences of spectral curves and peak spectral features in different bands of apple hyperspectral images before and after segmentation,and inverting the two-dimensional spatial distribution of apples and fruit stalks from a one-dimensional perspective;extracting apple spectral features by principal component analysis(PCA)and competitive adaptive reweighting algorithm(CARS),and establishing a damage detection model by combining genetic algorithm optimized support vector machine(GA-SVM)with an accuracy of 93.75%;the partial least squares(PLS)prediction model of apple sweetness and acidity was established by correlating spectral features and taste data of the electronic tongue using the continuous projection algorithm(SPA).The correlation between the spectral features and the sweetness and acidity at different wavebands was analyzed,and the optimal wavebands for sweetness and acidity detection were finally obtained as 532 nm and 624 nm,respectively.(2)Design and research of 3D reconstruction device.Designed and developed a threedimensional reconstruction device,specifying the method of image acquisition by the instrument;debugged and calibrated the RGB binocular camera to ensure the stability and accuracy of the device;made a characteristic band lens for sweetness and sourness,installed on the RGB binocular camera,and installed it on the RGB binocular camera.In the case of a specific light source,the lens filter is used,and the image data under a specific wavelength band is corrected,and finally the prediction result of the sweetness and sourness is visualized in a two-dimensional space.(3)Research on the reconstruction of apple 3D model and its quality rating.Use semiglobal matching(SGBM)and motion recovery structure(SFM)algorithms to obtain apple's point cloud data;calculate the obtained spatial point cloud envelope to estimate the true size of apple;calculate apple red based on color space and Hough gradient method The coloring rate and roundness are verified;the three-dimensional model of apple is obtained through multi-angle images,and the results of sour and sweetness are visualized in three-dimensional;The classification accuracy rate is above 80%.In this paper,we address the limitations of apple quality evaluation and the high cost of hyperspectral instruments and use hyperspectral image processing methods and 3D reconstruction technology to integrate and analyze and design a 3D reconstruction device that can adapt to multiple environments and visualize different physicochemical properties and precise geometric information of the detected objects in specific wavelengths to achieve a more comprehensive apple quality evaluation,and prove that the development of low-cost hyperspectral instruments It also demonstrates that the development of low-cost hyperspectral instruments has great application value.