Research on Waste Plastics Classification Method Based on Multi-Scale Feature Fusion.

Microplastic particles produced by non-degradable waste plastic bottles have a critical impact on the environment. Reasonable recycling is a premise that protects the environment and improves economic benefits. In this paper, a multi-scale feature fusion method for RGB and hyperspectral images based on Segmenting Objects by Locations (RHFF-SOLOv1) is proposed, which uses multi-sensor fusion technology to improve the accuracy of identifying transparent polyethylene terephthalate (PET) bottles, blue PET bottles, and transparent polypropylene (PP) bottles on a black conveyor belt. A line-scan camera and near-infrared (NIR) hyperspectral camera covering the spectral range from 935.9 nm to 1722.5 nm are used to obtain RGB and hyperspectral images synchronously. Moreover, we propose a hyperspectral feature band selection method that effectively reduces the dimensionality and selects the bands from 1087.6 nm to 1285.1 nm as the features of the hyperspectral image. The results show that the proposed fusion method improves the accuracy of plastic bottle classification compared with the SOLOv1 method, and the overall accuracy is 95.55%. Finally, compared with other space-spectral fusion methods, RHFF-SOLOv1 is superior to most of them and achieves the best (97.5%) accuracy in blue bottle classification.

Rapid dataset generation methods for stacked construction solid waste based on machine vision and deep learning.

The development of urbanization has brought convenience to people, but it has also brought a lot of harmful construction solid waste. The machine vision detection algorithm is the crucial technology for finely sorting solid waste, which is faster and more stable than traditional methods. However, accurate identification relies on large datasets, while the datasets from the field working conditions are scarce, and the manual annotation cost of datasets is high. To rapidly and automatically generate datasets for stacked construction waste, an acquisition and detection platform was built to automatically collect different groups of RGB-D images for instances labeling. Then, based on the distribution points generation theory and data augmentation algorithm, a rapid-generation method for synthetic construction solid waste datasets was proposed. Additionally, two automatic annotation methods for real stacked construction solid waste datasets based on semi-supervised self-training and RGB-D fusion edge detection were proposed, and datasets under real-world conditions yield better models training results. Finally, two different working conditions were designed to validate these methods. Under the simple working condition, the generated dataset achieved an F1-score of 95.98, higher than 94.81 for the manually labeled dataset. In the complicated working condition, the F1-score obtained by the rapid generation method reached 97.74. In contrast, the F1-score of the dataset obtained manually labeled was only 85.97, which demonstrates the effectiveness of proposed approaches.

RGB-D fusion models for construction and demolition waste detection.

The development of urbanization has brought a large amount of construction and demolition waste (CDW), which occupy land and cause adverse ecological effects. To effectively solve the negative impact of CDW, it needs to be recycled. Accurate waste classification is key to successful waste management. However, the current waste classification methods mainly use color images to classify, which cannot meet the needs of accurate classification. This paper built an RGB-depth (RGB-D) detection platform, using a color camera and a laser line-scanning sensor to collect RGB images and depth images. In order to use RGB images and depth images for feature fusion more effectively, this paper proposed three fusion models: RGB-D concat、RGB-D Ci-add and RGB-D Ci-concat. All these models based on an instance segmentation network called mask region convolutional neural network (Mask R-CNN), which can accurately segment the contours of each object while classifying them. The experimental results show that the mAPs of the RGB-D Ci-add / concat model are 1.33% to 1.72% higher than those of the RGB model, and the classification accuracy is 1.92% ∼ 2.27% higher. In addition, all the proposed models can meet the real-time requirement of online detection. Due to the excellent comprehensive performance of the RGB-D Ci-concat model, it can be regarded as the final detection model of the robot, which can improve the sorting efficiency of CDW further.

A robust classification algorithm for separation of construction waste using NIR hyperspectral system.

To improve the utilization rate of construction waste, reduce processing costs, and improve processing efficiency, we used near-infrared hyperspectral technology to extract and classify typical construction waste types. We proposed the pythagorean wavelet transform (PWT) to get the characteristic reflectivity to avoid the redundancy of hyperspectral data. Compared with the results from the wavelet transform (WT), we were able to retain more detailed information, and we observed the enhancement of differences between different species. To adapt to the complex conditions present in actual situations and to improve our ability to distinguish similar spectrum, we extracted, in addition to the characteristic reflectivity, four potential features. After classified verification, we found out that the first derivative and the intrinsic mode function (IMF) were effective features. At the same time the random forest (RF) algorithm was best at identifying trend-features, and the extreme learning machine (ELM) was better at identifying amplitude-features. We proposed a complementary troubleshooting (CT) method for the online identification of construction waste. After using the ELM to identify the characteristic reflectivity, the RF was used to identify first derivative for supplemental verification, which reduced errors due to working conditions and improved the overall model robustness and correctness. The accuracy of proposed method can reach 100% in identifying 180 samples with 6 types including woods, plastics, bricks, concretes, rubbers and black bricks.

Development of online classification system for construction waste based on industrial camera and hyperspectral camera.

Construction waste is a serious problem that should be addressed to protect environment and save resources, some of which have a high recovery value. To efficiently recover construction waste, an online classification system is developed using an industrial near-infrared hyperspectral camera. This system uses the industrial camera to capture a region of interest and a hyperspectral camera to obtain the spectral information about objects corresponding to the region of interest. The spectral information is then used to build classification models based on extreme learning machine and resemblance discriminant analysis. To further improve this system, an online particle swarm optimization extreme learning machine is developed. The results indicate that if a near-infrared hyperspectral camera is used in conjunction with an industrial camera, construction waste can be efficiently classified. Therefore, extreme learning machine and resemblance discriminant analysis can be used to classify construction waste. Particle swarm optimization can be used to further enhance the proposed system.

Detection of size of manufactured sand particles based on digital image processing.

The size distribution of manufactured sand particles has a significant influence on the quality of concrete. To overcome the shortcomings of the traditional vibration-sieving method, a manufactured sand casting/dispersing system was developed, based on the characteristics of the sand particle contours (as determined by backlit image acquisition) and an extraction mechanism. Algorithms for eliminating particles from the image that had be repeatedly captured, as well as for identifying incomplete particles at the boundaries of the image, granular contour segmentation, and the determination of an equivalent particle size, are studied. The hardware and software for the image-based detection device were developed. A particle size repeatability experiment was carried out on the single-grade sands, grading the size fractions of the manufactured sand over a range of 0.6-4.75 mm. A method of particle-size correction is proposed to compensate for the difference in the results obtained by the image-based method and those obtained by the sieving method. The experimental results show that the maximum repeatability error of single-grade fractions is 3.46% and the grading size fraction is 0.51%. After the correction of the image method, the error between the grading size fractions obtained by the two methods was reduced from 7.22%, 6.10% and 5% to 1.47%, 1.65%, and 3.23%, respectively. The accuracy of the particle-size detection can thus satisfy real-world measuring requirements.

An arc fault diagnosis algorithm using multiinformation fusion and support vector machines.

Arc faults in low-voltage electrical circuits are the main hidden cause of electric fires. Accurate identification of arc faults is essential for safe power consumption. In this paper, a detection algorithm for arc faults is tested in a low-voltage circuit. With capacitance coupling and a logarithmic detector, the high-frequency radiation characteristics of arc faults can be extracted. A rapid method for computing the current waveform slope characteristics of an arc fault provides another characteristic. Current waveform periodic integral characteristics can be extracted according to asymmetries of the arc faults. These three characteristics are used to develop a detection algorithm of arc faults based on multiinformation fusion and support vector machine learning models. The tests indicated that for series arc faults with single and combination loads and for parallel arc faults between metallic contacts and along carbonization paths, the recognition algorithm could effectively avoid the problems of crosstalk and signal loss during arc fault detection.