Deep learning-based image classification and quantification models for tablet sticking.

Super-resolution technique can be used to fix the low resolution problem for recognizing the iris at a distance. Two frequency domain super-resolution algorithms, Papoulis-Gerchberg (PG) and Projection onto Convex Sets, are implemented to increase the resolution of iris images. The performance analysis of these algorithms is carried out by extracting Gray Level Co-occurrence Matrix (GLCM) features of super-resoluted iris images. The super-resoluted iris region is normalized, extracted GLCM features and compared with the GLCM features of normalized original iris region. It has been observed that the GLCM features reconstructed images using above algorithm closely matches with that of original iris image. The error between the GLCM features of original normalized and normalized super-resoluted image using Papoulis-Gerchberg is less compared to that of Projection onto Convex Sets.

Breast cancer is a prevalent malignancy characterized by the uncontrolled growth of glandular epithelial cells, which can metastasize through the blood and lymphatic systems. Microcalcifications, small calcium deposits within breast tissue, are critical markers for early detection of breast cancer, especially in non-palpable carcinomas. These microcalcifications, appearing as small white spots on mammograms, are challenging to identify due to potential confusion with other tissues. This study hypothesizes that a hybrid feature extraction approach combined with Convolutional Neural Networks (CNNs) can significantly enhance the detection and localization of microcalcifications in mammograms. The proposed algorithm employs Gabor, Prewitt, and Gray Level Co-occurrence Matrix (GLCM) kernels for feature extraction. These features are input to a CNN architecture designed with maxpooling layers, Rectified Linear Unit (ReLU) activation functions, and a sigmoid response for binary classification. Additionally, the Top Hat filter is used for precise localization of microcalcifications. The preprocessing stage includes enhancing contrast using the Volume of Interest Look-Up Table (VOI LUT) technique and segmenting regions of interest. The CNN architecture comprises three convolutional layers, three ReLU layers, and three maxpooling layers. The training was conducted using a balanced dataset of digital mammograms, with the Adam optimizer and binary cross-entropy loss function. Our method achieved an accuracy of 89.56%, a sensitivity of 82.14%, and a specificity of 91.47%, outperforming related works, which typically report accuracies around 85-87% and sensitivities between 76 and 81%. These results underscore the potential of combining traditional feature extraction techniques with deep learning models to improve the detection and localization of microcalcifications. This system may serve as an auxiliary tool for radiologists, enhancing early detection capabilities and potentially reducing diagnostic errors in mass screening programs.

Global blindness is substantially influenced by age-related macular degeneration (AMD). It significantly shortens people's lives and severely impairs their visual acuity. AMD is becoming more common, requiring improved diagnostic and prognostic methods. Treatment efficacy and patient survival rates stand to benefit greatly from these upgrades. To improve AMD diagnosis in preprocessed retinal images, this study uses Grey Level Co-occurrence Matrix (GLCM) features for texture analysis. The selected GLCM features include contrast and dissimilarity. Notably, grayscale pixel values were also integrated into the analysis. Key factors such as contrast, correlation, energy, and homogeneity were identified as the primary focuses of the study. Various supervised machine learning (ML) and CNN techniques were employed on Optical Coherence Tomography (OCT) image datasets. The impact of feature selection on model performance is evaluated by comparing all GLCM features, selected GLCM features, and grayscale pixel features. Models using GSF features showed low accuracy, with OCTID at 23% and Kermany at 54% for BC, and 23% and 53% for CNN. In contrast, GLCM features achieved 98% for OCTID and 73% for Kermany in RF, and 83% and 77% in CNN. SFGLCM features performed the best, achieving 98% for OCTID across both RF and CNN, and 77% for Kermany. Overall, SFGLCM and GLCM features outperformed GSF, improving accuracy, generalization, and reducing overfitting for AMD detection. The Python-based research demonstrates ML's potential in ophthalmology to enhance patient outcomes.

Accurate detection of malignant tumor on lung computed tomography scans is crucial for early diagnosis of lung cancer and hence the faster recovery of patients. Several deep learning methodologies have been proposed for lung tumor detection, especially the convolution neural network (CNN). However, as CNN may lose some of the spatial relationships between features, we plan to combine texture features such as fractal features and gray-level co-occurrence matrix (GLCM) features along with the CNN features to improve the accuracy of tumor detection. Our framework has two advantages. First it fuses the advantage of CNN features with hand-crafted features such as fractal and GLCM features to gather the spatial information. Second, we reduce the overfitting effect by replacing the softmax layer with the support vector machine classifier. Experiments have shown that texture features such as fractal and GLCM when concatenated with deep features extracted from DenseNet architecture have a better accuracy of 95.42%, sensitivity of 97.49%, and specificity of 93.97%, and a positive predictive value of 95.96% with area under curve score of 0.95.

Background: Digital breast tomosynthesis (DBT) systems employ a sophisticated set of acquisition parameters to generate an image set, and the DBT acquisition angle is considered to be one of the most important parameters. The aim of this study was to use texture analysis to assess how the DBT acquisition angle might influence DBT images of breast parenchyma. Methods: Thirty-four patients were selected from a clinical study conducted at IRST Institute. Each patient underwent a dual DBT scan performed with Fujifilm Amulet Innovality (Fujifilm Corp, Tokyo, Japan) in standard (ST, angular range = 15°) and high-resolution (HR, angular range = 40°) modalities. Texture analysis was applied on the paired dataset using histogram-based features and gray level co-occurrence matrix (GLCM) features. Wilcoxon-signed rank and Pearson-rank tests were used to assess the statistical differences and correlations between extracted features. Results: The DBT acquisition angle did not affect histogram-based features, whereas there was a significant difference in five GLCM features (p < 0.05) between DBT images generated with 15° and 40° acquisition angles. Correlation analysis showed that two GLCM features were not correlated at a p < 0.05 significance level. Conclusions: DBT acquisition angle affects the textures extracted from DBT images and this dependence should be considered when establishing baselines for classifiers of malignant tissue. Furthermore, texture analysis could be proposed as a quantitative method for comparing and scoring the contrast of DBT images.

Batik has a vast variety of motifs and colors. Aside from its popularity as being part of Indonesian culture, it has become the source of Indonesia’s income. Batik was more promising in the past years for the business opportunities. Batik has economic and high export value as the commodity. Batik has become the main part of national culture; however there is a lack of understanding for many people, as they are still unaware about batik motifs and patterns. Therefore, it is needed for building a model to identify batik motifs. This study aims to combine the features of texture and the feature of shapes’ ornament in batik to classify images using artificial neural networks. The value of texture features of images in batik is extracted using gray level co-occurrence matrices (GLCM) which include Contrast, Correlation, Homogeneity and Energy. And include the Gray level Run length matrices (GLRLM) which includes Gray Level Non-Uniformity (GLN), Long Run Emphasis (LRE), Short Run Emphasis (SRE), Run Percentage (RP). At this phase of the training and testing, we compare the value of a classification accuracy of neural networks in each class in batik with their texture features, and the combination of GLCM and GLRLM. From the three features used in the classification of batik image with artificial neural networks it includes Probabilistic Neural network, it was obtained that GLCM feature has the lowest accuracy rate of 78% and the combination of GLCM and GLRLM features produces a greater value of accuracy by 84%. The results obtained in this study indicate that there is an increase in accuracy of batik image classification using the probabilistic neural network with the combination of GLCM and GLRLM features in batik image.

Compared to various optical remote sensing data, studies on the performance of dual-pol Synthetic aperture radar (SAR) on lithology discrimination are scarce. This study aimed at using Sentinel-1 data to distinguish dolomite, andesite, limestone, sandstone, and granite rock types. The backscatter coefficients VV and VH, the ratio VV–VH; the decomposition parameters Entropy, Anisotropy, and Alpha were firstly derived and the Kruskal–Wallis rank sum test was then applied to these polarimetric derived matrices to assess the significance of statistical differences among different rocks. Further, the corresponding gray-level co-occurrence matrices (GLCM) features were calculated. To reduce the redundancy and data dimension, the principal component analysis (PCA) was carried out on the GLCM features. Due to the limited rock samples, before the lithology discrimination, the input variables were selected. Several classifiers were then used for lithology discrimination. The discrimination models were evaluated by overall accuracy, confusion matrices, and the area under the curve-receiver operating characteristics (AUC-ROC). Results show that (1) the statistical differences of the polarimetric derived matrices (backscatter coefficients, ratio, and decomposition parameters) among different rocks was insignificant; (2) texture information derived from Sentinel-1 had great potential for lithology discrimination; (3) partial least square discrimination analysis (PLSDA) had the highest overall accuracy (0.444) among the classification models; (4) though the overall accuracy is unsatisfactory, according to the AUC-ROC and confusion matrices, the predictive ability of PLSDA model for limestone is high with an AUC value of 0.8017, followed by dolomite with an AUC value of 0.7204. From the results, we suggest that the dual-pol Sentinel-1 data are able to correctly distinguish specific rocks and has the potential to capture the variation of different rocks.

Cloud detection of satellite images is a challenging task. Extracting discriminative image features is one of the crucial steps for accurate cloud detection. In this chapter, we introduce a texture-based image feature that combines the merits of grey-level co-occurrence matrix (GLCM) features and rotation invariant uniform local binary pattern (RIULBP). The cloud detection method based on proposed feature consists of three steps: (1) Enhancing the image and dividing it into non-overlap patches; (2) Calculating GLCM features and RIULBP independently on patches and determining their optimal key parameters based on cloud detection performance; (3) Combining optimal GLCM features and RIULBP and feeding them into SVM classifier to identify patches with cloud. The proposed detection method is quantitatively compared to methods that only use GLCM features and RIULBP. The overall detection accuracy shows that our proposed method outperforms the GLCM and RIULBP method on real images. The proposed cloud detection method facilitates cloud segmentation and classification tasks, which can aid to better analysis of satellite image.

Introduction Phytoplankton are microscopic organisms that form the foundation of aquatic food webs. Accurate identification and classification of phytoplankton species are crucial for monitoring all aquatic ecosystems, from marine to freshwater, understanding ecological dynamics, and assessing environmental changes. Traditional methods of phytoplankton identification, which rely on manual microscopy, are time-consuming and require expert knowledge. Recent advancements in machine learning, particularly Convolutional Neural Networks (CNNs), offer promising solutions for automating this process. This abstract explores the application of pre-trained CNNs in recognizing phytoplankton species, highlighting their advantages, methodologies, and potential impacts. Methodology We present three approaches from a marine site, the Gulf of Venice site of the LTER-Italy network (DEIMS.ID https://deims.org/758087d7-231f-4f07-bd7e-6922e0c283fd), which includes the 'Acqua Alta' Oceanographic Tower (AAOT) (Fig. 1), the brackishwater site Utö Atmospheric and Marine Research Station (ResNet-18, located at 59°46.84’ N, 21°22.13’ E) https://en.ilmatieteenlaitos.fi/uto, and the freshwater site the IGB-LakeLab in Lake Stechlin NE Germany (DEIMS.ID https://deims.org/2223bc9c-12b2-49fe-af73-4299f553e054). Three different architectures of CNN were used: VGG16 for the Gulf of Venice, ResNet-18 for the Finnish station and a YOLOv11-cls for the German Lake Stechlin LakeLab station. These CNN models were pre-trained on the ImageNet dataset and subsequently fine-tuned with specific datasets for the respective geographic areas. These CNNs were chosen for their ability to autonomously extract features from images without external assistance, making them efficient, fast tools for analyzing large amounts of data and due to their specificity regarding the characteristics of the observational site. The process involves several steps: Data Collection and Preprocessing : several public datasets are available (Ciranni et al. 2024), where each image is annotated according to its class. Each model is structured to require input images in a specific format, so depending on the chosen model, it is necessary to preprocess the images accordingly. With an Imaging Flow Cytobot (IFCB, an in-situ automated submersible imaging flow cytometer that generates images of particles in-flow taken from the aquatic environment.), the produced images are of good quality (Fig. 2), and the main modification applied is resizing the images to fit the model requirements; Transfer Learning : Transfer learning allows the weights of a pre-trained neural network to be retained and updated (only if specified) for specific tasks. It has been demonstrated that using pre-trained models leads to significant results, reducing both training time and the amount of data required compared to an untrained model (Maracani et al. 2023); Training and Validation : The modified CNN is trained on the annotated phytoplankton images. Techniques such as data augmentation (to increment the number of images), dropout, and batch normalization are employed to enhance model performance and prevent overfitting. The model's accuracy is validated using a separate dataset; Evaluation Metrics : Performance metrics, including accuracy, precision, recall, and F1-score, are used to evaluate the model. Confusion matrices and receiver operating characteristic (ROC) curves provide additional insights into the model's classification capabilities. Data Collection and Preprocessing : several public datasets are available (Ciranni et al. 2024), where each image is annotated according to its class. Each model is structured to require input images in a specific format, so depending on the chosen model, it is necessary to preprocess the images accordingly. With an Imaging Flow Cytobot (IFCB, an in-situ automated submersible imaging flow cytometer that generates images of particles in-flow taken from the aquatic environment.), the produced images are of good quality (Fig. 2), and the main modification applied is resizing the images to fit the model requirements; Transfer Learning : Transfer learning allows the weights of a pre-trained neural network to be retained and updated (only if specified) for specific tasks. It has been demonstrated that using pre-trained models leads to significant results, reducing both training time and the amount of data required compared to an untrained model (Maracani et al. 2023); Training and Validation : The modified CNN is trained on the annotated phytoplankton images. Techniques such as data augmentation (to increment the number of images), dropout, and batch normalization are employed to enhance model performance and prevent overfitting. The model's accuracy is validated using a separate dataset; Evaluation Metrics : Performance metrics, including accuracy, precision, recall, and F1-score, are used to evaluate the model. Confusion matrices and receiver operating characteristic (ROC) curves provide additional insights into the model's classification capabilities. Results Studies have demonstrated that pre-trained CNNs can achieve high accuracy in phytoplankton classification. In our case, models like ResNet and VGG have shown classification accuracies exceeding 80% on diverse phytoplankton datasets (Fig. 3, Kraft et al. 2022). These models effectively distinguish between species with subtle morphological differences, which are often challenging for human experts. Discussion The use of pre-trained CNNs in phytoplankton recognition offers several advantages: Efficiency : Automated classification significantly reduces the time and effort required for phytoplankton identification compared to manual methods. Scalability : CNNs can handle large volumes of image data, making them suitable for Long Term Ecological Research. Consistency : Machine learning models provide consistent and objective classifications, minimizing human error and variability. Efficiency : Automated classification significantly reduces the time and effort required for phytoplankton identification compared to manual methods. Scalability : CNNs can handle large volumes of image data, making them suitable for Long Term Ecological Research. Consistency : Machine learning models provide consistent and objective classifications, minimizing human error and variability. However, challenges remain. The automatic taxonomic identification level is still not as detailed as that of human expertise. The quality and diversity of training data are critical for model performance. Inadequate or biased datasets can lead to poor generalization. Additionally, the interpretability of CNNs is limited, making it difficult to understand the decision-making process fully. Conclusion Pretrained CNNs represent a powerful tool and a pipeline for phytoplankton species recognition, offering significant improvements in efficiency, scalability, and consistency over traditional methods. Continued advancements in machine learning and the availability of high-quality datasets will further enhance the capabilities of these models. Future research should focus on addressing current limitations, such as data quality and model interpretability, to fully realize the potential of CNNs in marine science. In this work, we will present the results as discussed to demonstrate possible workflows to fully realize the potential of CNNs in marine science and potentially contribute to the Standard Observations (SOs) addressing current limitations. We will also bring a workflow proposal to manage and perform actions related to harmonization, interoperability, quality control and sharing of the data obtained througth the CNNs recognitions following the directives proposed by Torstensson (2025).

Lasem batik is one of traditional Indonesian’s famous batik that has high artistic and economic value. Lasem batik can be classified into several different motifs. This study aims to analyze the number of gray level co-occurrence matrix (GLCM) features during the process of classification of Lasem batik image with K-Nearest Neighbor (KNN). KNN has advantages in overcoming probability density, able to consolidate calculations based on the number of neighbors specified, and can perform calculations with limited parameters. Feature extraction is one of the important steps before performing image classification. GLCM is one of the extraction features of a very popular texture. There are five GLCM features that are widely used, namely contrast, homogeneity, energy, correlation, and entropy. This research classifies five kinds of the famous motif of batik Lasem. The training and testing process compare the use of four and five GLCM features for each of the three experiments with different amounts of data. The test results show that the use of four and five types of GLCM features get the same accuracy in each experiment. It can be concluded that with KNN enough to use four kinds of features to speed up the calculation of the classification.

Signature is widely used and developed area of research for personal verification and authentication. In this paper, we present a new offline handwritten signature recognition system based on fusion of global and GLCM (Grey Level Co-occurrence Matrix) features using fuzzy logic system as classifier tool. The global and GLCM features are fused to generate vector of 15 features for the verification of the signature. The test signature is compared with the database signatures based on features, whilst match/non match of signatures is decided with fuzzy logic. The experimental results obtained by using a database of 7 individuals’ signatures. A total number of 70 images are collected for our study and with average 10 signatures for each person, 5 of the signatures are used as training, the remaining 5signatures are used as testing group. The results show that the proposed modular architecture can achieve 100% recognition accuracy for training group and 90.5% recognition accuracy for the testing group with running time is 1.17 second.

PURPOSE Triple negative breast cancer (TNBC) is an aggressive and heterogeneous subtype of breast cancer. Pathologic complete response (pCR) to neoadjuvant systemic therapy (NAST) predicts better survival. Early prediction of the treatment response can potentially triage non-responding patients to alternative protocol treatments, spare them of the unneeded toxicity, and improve pCR. We evaluated the ability of radiomic textural analysis of intratumoral and peritumoral regions on the dynamic contrast enhanced (DCE) and diffusion-weighted imaging (DWI) MRI images obtained early during NAST to predict pCR. MATERIALS AND METHODS This IRB-approved prospective study (NCT02276443) included 182 patients with biopsy proven stage I-III TNBC who had multiparametric MRIs at baseline (BL), post 2 cycles (C2), and post 4 cycles (C4) of NAST before surgery. Tumors and peritumoral regions of 5 mm and 10 mm in thickness were segmented on the 2.5 minutes DCE subtraction images and on the b=800 DWI images. Ten histogram-based first order texture features including mean, minimum, maximum, standard deviation, kurtosis, skewness, 1st, 5th, 95th, and 99th percentile, and 300 radiomic Grey Level Co-occurrence matrix (GLCM) features along with their absolute and relative differences between the 3 imaging time points were extracted from the tumors and from the peritumoral regions with an in-house Matlab toolbox. Treatment response at surgery (pCR vs non-pCR) was documented. The samples were divided into training and testing datasets by a 2:1 ratio. Area under the receiver operating characteristics curve (AUC ROC) was calculated for univariate analysis in predicting pCR. Logistic regression with elastic net regularization was performed for texture feature selection. Parameter optimization was performed by using 5-fold cross-validation based on mean cross-validated AUC in the training set. RESULTS Of 182 TNBC patients, 88 (48%) had pCR and 94 (52%) did not achieve pCR. Eight multivariate models combining radiomic features from both DCE and DWI tumoral and peritumoral regions had AUC > 0.8 (0.807-0.831) with p-value < 0.001 in both training and testing sets. The highest AUC=0.831 was obtained from a model consisting of 15 radiomic features: tumor DWI (5 GLCM features) at C2, peritumoral region on DCE (skewness) at C2, tumor DCE (1st, 5th percentile) at C4, tumor DWI (3 GLCM features) at C4, peritumoral region DWI (1 GLCM feature) at C4, and the relative difference between C4/C2 on DCE (5th, 95th percentile and mean). CONCLUSION Multi-parametric MRI-based radiomics models from the tumor and the peritumoral regions showed high accuracy as potential early predictors of NAST response in TNBC patients. Citation Format: Rania M. Mohamed, Bikash Panthi, Beatriz Adrada, Rosalind Candelaria, Mary S. Guirguis, Wei Yang, Medine Boge, Miral Patel, Nabil Elshafeey, Sanaz Pashapoor, Zijian Zhou, Jong Bum Son, Ken-Pin Hwang, H. T. Carisa Le-Petross, Jessica Leung, Marion E. Scoggins, Gary J. Whitman, Zhan Xu, Deanna L. Lane, Tanya Moseley, Frances Perez, Jason White, Elizabeth Ravenberg, Alyson Clayborn, Mark Pagel, Huiqin Chen, Jia Sun, Peng Wei, Alastair M. Thompson, Stacy Moulder, Anil Korkut, Lei Huo, Kelly K. Hunt, Jennifer K. Litton, Vicente Valero, Debu Tripathy, Clinton Yam, Jingfei Ma, Gaiane Rauch. Multi-Parametric MRI-Based Radiomics Models from Tumor and Peritumoral Regions as Potential Predictors of Treatment Response to Neoadjuvant Systemic Therapy in Triple Negative Breast Cancer Patients [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P6-01-06.

From staying quarantined at home, practicing work from home to moving outside wearing masks and carrying sanitizers, every individual has now become so adaptive to so called ‘New Normal’ post series of lockdowns across the countries. The situation triggered by novel Coronavirus has changed the behaviour of every individual towards every other living as well as non-living entity. In the Wuhan city of China, multiple cases were reported of pneumonia caused due to unknown reasons. The concerned medical authorities confirmed the cause to be Coronavirus. The symptoms seen in these cases were not much different than those seen in case of pneumonia. Earlier the research has been carried out in the field of pneumonia identification and classification through X-ray images of chest. The difficulty in identifying Covid19 infection at initial stage is due to high resemblance of its symptoms with the infection caused due to pneumonia. Hence it is trivial to well distinguish cases of coronavirus from pneumonia that may help in saving life of patients. The paper uses chest X-ray images to identify Covid19 infection in lungs using machine learning classifiers and ensembles with Gray-Level Cooccurrence Matrix (GLCM) features. The advocated methodology extracts statistical texture features from X-ray images by computing a GLCM for each image. The matrix is computed by considering various stride combinations. These GLCM features are used to train the machine learning classifiers and ensembles. The paper explores both the multiclass classification (X-ray images are classified into one of the three classes namely Covid19 affected, Pneumonia affected and normal lungs) and binary classification (Covid19 affected and other). The dataset used for evaluating performance of the method is open sourced and can be accessed easily. Proposed method being simple and computationally effective achieves noteworthy performance in terms of Accuracy, F-Measure, MCC, PPV and Sensitivity. In sum, the best stride combination of GLCM and ensemble of machine learning classifiers is suggested as vital outcome of the proposed method for effective Covid19 identification from chest X-ray images

Objective: To design a computer aided detection system for the early detection of the breast cancer from the mammograms as it can assist the doctors in the diagnosis. Methodology: The proposed method used in the design of Computer-Aided Detection (CAD) is based the on using the textural differences of the abnormal and normal mammograms to detect the breast cancer. The Gabor Features and gray-level co-occurrence matrix (GLCM) features are extracted from the region of interest of the segmented mammograms using the Entropy based segmentation. The Support Vector Machine (SVM) classifier is used for classifying the mammograms into the cancerous and non-cancerous cases. The 35 number of normal and abnormal mammograms are taken from the Mammographic Image Analysis Society (MIAS) data set. The MIAS database is chosen as it carries more challenging data because it carries lot of unwanted tissues and flesh part included in it which has the intensity level more than the micro-calcifications. Findings: The classification accuracies obtained are 92.98% and 98.11% using Gabor and gray-level co-occurrence matrix features respectively. The sensitivity achieved with the gray-level co-occurrence matrix features is 100% which shows no missed cancerous case. The classification accuracy is higher using gray-level co-occurrence matrix features as compared to the Gabor features which shows superiority of these features in capturing the texture of the mammograms. Novelty: The removal of the pectoral muscle is an important pre-processing step. Only with the proper elimination of the pectoral muscle, the segmentation of the mammograms is possible. The method proposed to remove the pectoral muscle in this paper removed the pectoral muscles from all the mammograms used in the study. The entropy based segmentation and the technique of the removal of the non-contributing features outperforms other CAD systems in the literature available. These are very promising results for successful design of a Computer-aided detection system for early detection of the breast cancer that can be put to clinical trials or used for the double reading. Keywords: Breast Cancer; Computer Aided Detection System; Entropy based Segmentation; Gabor Features; GrayLevel CoOccurrence Matrix Features

In this paper, the gray level co-occurrence matrix (GLCM) and histogram of oriented gradient (HOG) features fusion of time-frequency image are introduced into the reciprocating compressor fault diagnosis. Firstly, vibration signals are acquired from the reciprocating compressor in different states of head tile and the wavelet transform distributions of vibration signals were displayed in time-frequency images. Secondly, GLCM and HOG methods are used to extract features from time-frequency images, then GLCM feature and HOG feature are fused and input into support vector machine for recognition and classification. By this way, the fault diagnosis of time series signals of reciprocating compressor is transferred to the classification of time-frequency images. The results show that can accurately realize diagnosis of small-head wear fault of reciprocating compressor.