Advanced Image Processing Methods Research Articles

A Critical Evaluation of Optimizing the Potential for Ore Recovery from Mine Waste Using Multi-Elemental, Geochemical and Petrographic Analysis (MEGPA)

The growing world demand for metals necessitates the economic extraction of metals from ores of lesser grades and scheduled waste with potential for metal recovery. In which case, efficient skilled-beneficiation is required to achieve such demand. This research paper examines the potential of ore recovery from magmatic Ni-Cu-PGE waste rock in an open pit mine operation. The waste material contains a marginal cut-off grade between the ore and what has been scheduled to be produced as waste throughout its life of mine (LOM). The waste material has the potential for metal recovery to extend the LOM. The main aim was to investigate the appropriateness of the potential for metal recovery using the already generated waste as a baseline model for subsequent waste production. To achieve this aim, the primary objective was to use Multi-elemental, Geochemical and Petrographic Analysis (MEGPA) on carefully selected waste rock samples. While the specific objectives were to examine the liberation of metals and establish optimal cut-off grade based on suitable size fractions for ore recovery. Multi-elemental, geochemical and Petrographic Analysis were carried out on selected size fractions using various instruments such as Scanning Electron Microscope Energy Disperse X-ray (SEM-EDX), X-Ray Diffraction (XRD) and X-Ray Fluorescence (XRF), reflected and transmitted microscopes. Categorization of results shown the predominant bulk mineral abundance to be tremolite, a member of the amphibole group of silicate minerals with composition Ca2Si8O22·(OH)2. Moreover, the result revealed that the economic Ni and Cu sulphides are hosted in pentlandite and chalcopyrite within the waste samples. Liberation of mineral and recovery of metal are in the finer fractions size below 1.5 mm, and that optimum metal recovery is at ≤1 mm. It was concluded from the study that, there is potential for metal recovery from mines waste in finer fractions below 1.5 mm without posing significant metalliferous acid mine drainage (AMD) risk to the mining environment. It is recommended that further study about the influence on the metallurgical processing of the waste at the selected optimum fractions size, and energy requirement for screening the sample is carried out. Since the PSD (particle size distribution) is critical to both the physical separation processes of the waste and the prediction of metalliferous AMD risk, other methods like advanced UAV 3D photogrammetry and digital image processing method could be used to test for both underestimation and overestimation of PSD. This is important as the amount of mineral liberation, metal recovery and sulphur generation are PSD dependent.

Real-Time Thermal Imaging-Based System for Asphalt Pavement Surface Distress Inspection and 3D Crack Profiling

Infrared thermography is a cost-effective nonintrusive testing approach to assess surface and near-surface distresses, such as asphalt pavement surface cracks. However, the raw data collected by thermal cameras cannot be directly used for pavement surface distress inspection. Thus, advanced thermal image processing methods are desirable for extracting indicators of hidden flaws. The objective of this research was to develop an integrated system that combines infrared imaging, high-resolution visual imaging, real-time image processing, and data-rich analytics for automated inspection to support decision making for pavement preservative maintenance. This work developed a code that can integrate the characteristics captured by both thermal and visual images to provide a quantitative identification of the surface cracks and their severity. Field-test data were collected and a statistical analysis was conducted to correlate temperature gradient to the surface crack profile of asphalt pavement. It was found that the surface temperature distribution pattern has a direct correlation with pavement crack profile, and can be used as an indicator of crack depth. The proposed real-time thermal imaging-based system was found to be feasible for field inspection, and the three-dimensional (3D) crack profiling method can help produce fairly accurate measurements to enable fast and easy decision support for pavement preservation practices.

Feature extraction for hyperspectral mineral domain mapping: A test of conventional and innovative methods

Hyperspectral (HS) imaging holds great potential for the mapping of geological targets. Innovative acquisition modes such as drone-borne or terrestrial remote sensing open up new scales and angles of observation, which allow to analyze small-scale, vertical, or difficult-to-access outcrops. A variety of available sensors operating in different spectral ranges can provide information about the abundance and spatial location of various geologic materials. However geological outcrops are inherently uneven and spectrally heterogeneous, may be covered by dust, lichen or weathering crusts, or contain spectrally indistinct objects, which is why classifications or domain mapping approaches are often used in geoscientific and mineral exploration applications as a means to discriminate mineral associations (e.g. ore or alteration zones) based on overall variations in HS data. Feature extraction (FE) algorithms are prominently used as a preparatory step to identify the first order variations within the data and, simultaneously, reduce noise and data dimensionality. The most established FE algorithms in geosciences are, by far, Principal Component Analysis (PCA) and Minimum Noise Fraction (MNF). Major progress has been conducted in the image processing community within the last decades, yielding innovative FE methods that incorporate spatial information for smoother and more accurate classification results. In this paper, we test the applicability of conventional (PCA, MNF) and innovative FE techniques (OTVCA: Orthogonal total variation component analysis and WSRRR: Wavelet-based sparse reduced-rank regression) on three case studies from geological HS mapping campaigns, including drone-borne mineral exploration, terrestrial paleoseismic outcrop scanning and thermal HS lithological mapping. This allows us to explore the performance of different FE approaches on complex geological data with sparse or partly inaccurate validation data. For all case studies, we demonstrate advantages of innovative FE algorithms in terms of classification accuracy and geological interpretability. We promote the use of advanced image processing methods for applications in geoscience and mineral exploration as a tool to support geological mapping activities.

Mathematical modeling of unmanned movement aircraft - four-rotor

The article presents an analytical approach to building a mathematical model of a quadrocopter. The main purpose of building the model was to design an appropriate facility control system and analyze its behavior in various situations. The assumption was made to build a model, control system and all accompanying algorithms in an open programming environment, which will allow their subsequent implementation in a real facility, without the need to use expensive software. The quadrocopter is controlled by the operator by means of hand movements that are read by the camera and properly interpreted using advanced image processing methods. The entire system is visualized and embedded in a three-dimensional simulation environment. The model study was conducted using a DC motor as an input data source. The operation of the model was checked with a controller when a disturbance was introduced into the model. The four-rotor model with a selected regulator was tested by analyzing the angular velocity and position of the object in a rectangular coordinate system. At the end of the article, the results of the simulations made are presented and the resulting conclusions are presented.

Examination on Advanced Machine Learning Techniques for Plant Leaf Disease Detection from Leaf Imagery

Agriculture is the backbone of countries as it feeds people across the globe. With respect to agriculture in India, the total geographical area for agriculture in India is 328 million hectares. The new area sown is 142 million hectares. 190.8 million hectares is the gross cropped area. It reveals the significance of agriculture and its related research. With technological innovations emerged, now it is the era where there is scope of increasing the research towards precision agriculture (PA). PA is the agriculture where technology-driven procedures are followed to do right things at right time. This will increase productivity and saves time, money and effort. Agricultural plat leaf disease detection is an important research area in agriculture that is related to PA. In this context, image processing and machine learning techniques assume significance to capture live leave imagery from fields and detect diseases. Automatic detection of disease based on agricultural plant leaves has its utility in real world agricultural applications. This is the motivation behind taking up this research. This paper throws light on different kinds of plant diseases, various advanced machine learning and image processing methods to detect plant diseases. The innovations in technologies lead to automatic detection of diseases even from remote places and take necessary steps to ensure that the agricultural crops are healthy and yield desired harvest. This survey also provides important research gaps that will help in further research towards realizing PA.

Damage evolution of PµLSE additive-manufactured micro-lattice metastructures: Synchrotron radiation 3D tomography image-based analysis

The manufacturing of additives with projection micro litho stereo exposure (PμLSE) has provided an opportunity for the fabrication of metastructures with complex microstructures at micro-nano resolutions. However, the performance evaluation of as-fabricated metastructures is challenging. The benefit of synchrotron radiation-based 3D imaging techniques and advanced image processing methods makes it is feasible to study fabrication defects and damage processes of micro-nanoscale body-centered cubic (BCC) lattices manufactured with PμLSE. First, synchrotron radiation technology is used to capture the structural features inside the micro-lattice samples. Subsequently, several types of statistical defects-based image finite element models are adopted to analyze the failure process of the structure under compression loading. Finally, comparisons between in situ experiments and numerical simulation results are performed for verification. The method of the combined non-destructive testing of synchrotron radiation and image finite element technology provides a robust technique for evaluating the performances of additive-manufactured micro-lattice with complex microstructures.

Advanced Methods for Point Cloud Processing and Simplification

Nowadays, mobile robot exploration needs a rangefinder to obtain a large number of measurement points to achieve a detailed and precise description of a surrounding area and objects, which is called the point cloud. However, a single point cloud scan does not cover the whole area, so multiple point cloud scans must be acquired and compared together to find the right matching between them in a process called registration method. This method requires further processing and places high demands on memory consumption, especially for small embedded devices in mobile robots. This paper describes a novel method to reduce the burden of processing for multiple point cloud scans. We introduce our approach to preprocess an input point cloud in order to detect planar surfaces, simplify space description, fill gaps in point clouds, and get important space features. All of these processes are achieved by applying advanced image processing methods in combination with the quantization of physical space points. The results show the reliability of our approach to detect close parallel walls with suitable parameter settings. More importantly, planar surface detection shows a 99% decrease in necessary descriptive points almost in all cases. This proposed approach is verified on the real indoor point clouds.

Highlighting Directional Reflectance Properties of Retinal Substructures From D-OCT Images.

Optical coherence tomography (OCT), which is routinely used in ophthalmology, enables transverse optical imaging of the retina and, hence, the identification of the different neuronal layers. Directional OCT (D-OCT) extends this technology by acquiring sets of images at different incidence angles of the light beam. In this way, reflectance properties of photoreceptor substructures are highlighted, enabling physicians to study their orientation, which is potentially an interesting biomarker for retinal diseases. Nevertheless, commercial OCT devices equipped to automate D-OCT acquisition do not yet exist, meaning that physicians manually deviate the light beam to acquire a set of D-OCT images sequentially. Therefore, the intensities in the stack of images are not directly comparable, and a normalization step is required before differential analysis. In this paper, we present advanced image processing methods to perform differential analysis of a set of D-OCT images and extract the angle-dependent retinal substructures. Our approach relies on a robust and accurate normalization algorithm followed by a classification that is spatially regularized. We also propose a robust color representation that facilitates interpretation of D-OCT data in general, by detecting and highlighting angle-dependent structures in healthy and diseased eyes. Experimental results show evidence of photoreceptor disarray in a variety of retinal diseases, demonstrating the potential medical interest of the approach.

Segmentation-Based Blood Flow Parameter Refinement in Cerebrovascular Structures Using 4-D Arterial Spin Labeling MRA.

Cerebrovascular diseases are one of the main global causes of death and disability in the adult population. The preferred imaging modality for the diagnostic routine is digital subtraction angiography, an invasive modality. Time-resolved three-dimensional arterial spin labeling magnetic resonance angiography (4D ASL MRA) is an alternative non-invasive modality, which captures morphological and blood flow data of the cerebrovascular system, with high spatial and temporal resolution. This work proposes advanced medical image processing methods that extract the anatomical and hemodynamic information contained in 4D ASL MRA datasets. A previously published segmentation method, which uses blood flow data to improve its accuracy, is extended to estimate blood flow parameters by fitting a mathematical model to the measured vascular signal. The estimated values are then refined using regression techniques within the cerebrovascular segmentation. The proposed method was evaluated using fifteen 4D ASL MRA phantoms, with ground-truth morphological and hemodynamic data, fifteen 4D ASL MRA datasets acquired from healthy volunteers, and two 4D ASL MRA datasets from patients with a stenosis. The proposed method reached an average Dice similarity coefficient of 0.957 and 0.938in the phantom and real dataset segmentation evaluations, respectively. The estimated blood flow parameter values are more similar to the ground-truth values after the refinement step, when using phantoms. A qualitative analysis showed that the refined blood flow estimation is more realistic compared to the raw hemodynamic parameters. The proposed method can provide accurate segmentations and blood flow parameter estimations in the cerebrovascular system using 4D ASL MRA datasets. The information obtained with the proposed method can help clinicians and researchers to study the cerebrovascular system non-invasively.

Spectral classification of sparse photon depth images

By illumination of target scenes using a set of different wavelengths, we demonstrate color classification of scenes, as well as depth estimation, in photon-starved images. The spectral signatures are classified with a new advanced statistical image processing method from measurements of the same scene, in this case using combinations of 33, 16, 8 or 4 different wavelengths in the range 500 - 820 nm. This approach makes it possible to perform color classification and depth estimation on images containing as few as one photon per pixel, on average. Compared to single wavelength imaging, this approach improves target discrimination by extracting more spectral information, which, in turn, improves the depth estimation since this approach is robust to changes in target reflectivity. We demonstrate color classification and depth profiling of complex targets at average signal levels as low as 1.0 photons per pixel from as few as 4 different wavelength measurements.

Effects of Sulfur Content and Ash Content in Lubricating Oil on the Aggregate Morphology and Nanostructure of Diesel Particulate Matter

Five different lubricating oils with three different sulfur contents (0.182%, 0.583%, and 1.060%) and three different ash contents (0.48%, 1.21%, and 1.99%) were used in this study to investigate their effects on the morphology of diesel particulate matter (PM). The aggregate morphology and nanostructure of primary particles in diesel PM were analyzed using transmission electron microscopy. The primary morphological parameters were determined using an advanced image-processing method. The results show that, with increasing sulfur content and ash content, the macroscopic structure of the primary particles can transform from chainlike to more complex and disordered agglomerated structures, whereas the inner cores change little, the shell thickness decreases, and the thickness of the amorphous materials attached to the outer shell increases. The fractal dimension (Df) of the PM aggregate morphology varies from 1.656 to 1.811, and the mean primary-particle diameter (dp) of the PM is mainly in the range of 10–...

A Computer-Based Brain Tumor Detection Approach with Advanced Image Processing and Probabilistic Neural Network Methods

For years, automation in tumor detection area has been commonly required because there might be a shortage of skilled radiologists in this area. In this study, a computer based brain tumor detection approach in magnetic resonance imaging (MRI) is proposed. The purpose of this study is to detect and to localize the tumor areas in the brain with using advanced image processing techniques and probabilistic neural network (PNN) method. For this study, firstly, a special clustering technique is used on the MRI images. Secondly, with using thresholding and level-set segmentation, tumor areas are detected successfully. Thirdly, with using the obtained tumor areas, automatic brain tumor classification and analysis are achieved with using PNN method. In the experimental part of the study, 25 neuroimages are used to optimize our system and 25 out-of-sample neuroimages are also used to test the approach. The preliminary diagnostic results can demonstrate the high classification accuracy for the image processing and the neural network structures.

Transfer learning on fused multiparametric MR images for classifying histopathological subtypes of rhabdomyosarcoma

This paper presents a deep-learning-based CADx for the differential diagnosis of embryonal (ERMS) and alveolar (ARMS) subtypes of rhabdomysarcoma (RMS) solely by analyzing multiparametric MR images. We formulated an automated pipeline that creates a comprehensive representation of tumor by performing a fusion of diffusion-weighted MR scans (DWI) and gadolinium chelate-enhanced T1−weighted MR scans (MRI). Finally, we adapted transfer learning approach where a pre-trained deep convolutional neural network has been fine-tuned based on the fused images for performing classification of the two RMS subtypes. We achieved 85% cross validation prediction accuracy from the fine-tuned deep CNN model. Our system can be exploited to provide a fast, efficient and reproducible diagnosis of RMS subtypes with less human interaction. The framework offers an efficient integration between advanced image processing methods and cutting-edge deep learning techniques which can be extended to deal with other clinical domains that involve multimodal imaging for disease diagnosis.

Image analysis driven single-cell analytics for systems microbiology

BackgroundTime-lapse microscopy is an essential tool for capturing and correlating bacterial morphology and gene expression dynamics at single-cell resolution. However state-of-the-art computational methods are limited in terms of the complexity of cell movies that they can analyze and lack of automation. The proposed Bacterial image analysis driven Single Cell Analytics (BaSCA) computational pipeline addresses these limitations thus enabling high throughput systems microbiology.ResultsBaSCA can segment and track multiple bacterial colonies and single-cells, as they grow and divide over time (cell segmentation and lineage tree construction) to give rise to dense communities with thousands of interacting cells in the field of view. It combines advanced image processing and machine learning methods to deliver very accurate bacterial cell segmentation and tracking (F-measure over 95%) even when processing images of imperfect quality with several overcrowded colonies in the field of view. In addition, BaSCA extracts on the fly a plethora of single-cell properties, which get organized into a database summarizing the analysis of the cell movie. We present alternative ways to analyze and visually explore the spatiotemporal evolution of single-cell properties in order to understand trends and epigenetic effects across cell generations. The robustness of BaSCA is demonstrated across different imaging modalities and microscopy types.ConclusionsBaSCA can be used to analyze accurately and efficiently cell movies both at a high resolution (single-cell level) and at a large scale (communities with many dense colonies) as needed to shed light on e.g. how bacterial community effects and epigenetic information transfer play a role on important phenomena for human health, such as biofilm formation, persisters’ emergence etc. Moreover, it enables studying the role of single-cell stochasticity without losing sight of community effects that may drive it.

Experimental study of bubbly flow using image processing techniques

This paper presents an experimental study of bubbly flows at relatively high void fractions using an advanced image processing method. Bubble overlapping is a common problem in such flows and the past studies often treat the overlapping bubbles as a whole, which introduces considerable measurement uncertainties. In this study, a hybrid method combining intersection point detection and watershed segmentation is used to separate the overlapping bubbles. In order to reconstruct bubbles from separated segments, a systematic procedure is developed which can preserve more features captured in the raw image compared to the simple ellipse fitting method. The distributions of void fraction, interfacial area concentration, number density and velocity are obtained from the extracted bubble information. High-speed images of air-water bubbly flows are acquired and processed for eight test runs conducted in a 30mm×10mm rectangular channel. The developed image processing scheme can effectively separate overlapping bubbles and the results compare well with the measurements by the gas flow meter and double-sensor conductivity probe. The development of flows in transverse and mainstream directions are analyzed and compared with the prediction made by the one-dimensional interfacial area transport equation (IATE) and the bubble number density transport equation.

Gastric Cancer Regional Detection System.

In this study, a novel system was created to localize cancerous regions for stomach images which were taken with computed tomography(CT). The aim was to determine the coordinates of cancerous regions which spread in the stomach area in the color space with using this system. Also, to limit these areas with a high accuracy ratio and to feedback to the user of this system were the other objectives. This integration was performed with using energy mapping, analysis methods and multiple image processing methods and the system which was consisted from these advanced algorithms was appeared. For this work, in the range of 25-40 years and when gender discrimination was insignificant, 30 volunteer patients were chosen. During the formation of the system, to exalt the accuracy to the maximum level, 2 main stages were followed up. First, in the system, advanced image processing methods were processed between each other and obtained data were studied. Second, in the system, FFT and Log transformations were used respectively for the first two cases, then these transformations were used together for the third case. For totally three cases, energy distribution and DC energy intensity analysis were done and the performance of this system was investigated. Finally, with using the system's unique algorithms, a non-invasive method was achieved to detect the gastric cancer and when FFT and Log transformation were used together, the maximum success rate was obtained and this rate was calculated as 83,3119 %.

An advanced image processing method to improve the spatial resolution of ion radiographies

We present an optimization method to improve the spatial resolution and the water equivalent thickness (WET) accuracy of ion radiographies. The method is designed for imaging systems measuring for each actively scanned beam spot the lateral position of the pencil beam and at the same time the Bragg curve (behind the target) in discrete steps without relying on tracker detectors to determine the ion trajectory before and after the irradiated volume. Specifically, the method was used for an imaging set-up consisting of a stack of 61 parallel-plate ionization chambers (PPIC) interleaved with absorber plates of polymethyl methacrylate (PMMA) working as a range telescope.The method uses not only the Bragg peak position, but approximates the entire measured Bragg curve as a superposition of differently shifted Bragg curves. Their relative weights allow to reconstruct the distribution of thickness around each scan spot of a heterogeneous phantom.The approach also allows merging the ion radiography with the geometric information of a co-registered x-ray radiography in order to increase its spatial resolution. The method was tested using Monte Carlo simulated and experimental proton radiographies of a PMMA step phantom and an anthropomorphic head phantom. For the step phantom, the effective spatial resolution was found to be 6 and 4 times higher than the nominal resolution for the simulated and experimental radiographies, respectively. For the head phantom, a gamma index was calculated to quantify the conformity of the simulated proton radiographies with a digitally reconstructed radiography (DRR) obtained from an x-ray CT and properly converted into WET. For a distance-to-agreement (DTA) of 2.5 mm and a relative WET difference (RWET) of 2.5%, the passing ratio was 100%/85% for the optimized/non-optimized case, respectively. When the optimized proton radiography was merged with the co-registered DRR, the passing ratio was 100% at DTA = 1.3 mm and RWET = 1.3%. A special interpolation method allows to strongly reduce the dose by using a coarser grid of the measured beam spot position with a 5 times larger grid distance. We show that despite a dose reduction of 25 times (leading to a dose of 0.016 mGy for the current imaging set-up), the image quality of the optimized radiographies remains fairly unaffected for both the simulated and experimental case.

An efficient total variation algorithm for super-resolution in fetal brain MRI with adaptive regularization

Although fetal anatomy can be adequately viewed in new multi-slice MR images, many critical limitations remain for quantitative data analysis. To this end, several research groups have recently developed advanced image processing methods, often denoted by super-resolution (SR) techniques, to reconstruct from a set of clinical low-resolution (LR) images, a high-resolution (HR) motion-free volume. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has been quite attracted by Total Variation energies because of their ability in edge preserving but only standard explicit steepest gradient techniques have been applied for optimization. In a preliminary work, it has been shown that novel fast convex optimization techniques could be successfully applied to design an efficient Total Variation optimization algorithm for the super-resolution problem. In this work, two major contributions are presented. Firstly, we will briefly review the Bayesian and Variational dual formulations of current state-of-the-art methods dedicated to fetal MRI reconstruction. Secondly, we present an extensive quantitative evaluation of our SR algorithm previously introduced on both simulated fetal and real clinical data (with both normal and pathological subjects). Specifically, we study the robustness of regularization terms in front of residual registration errors and we also present a novel strategy for automatically select the weight of the regularization as regards the data fidelity term. Our results show that our TV implementation is highly robust in front of motion artifacts and that it offers the best trade-off between speed and accuracy for fetal MRI recovery as in comparison with state-of-the art methods.

WE‐EF‐303‐04: An Advanced Image Processing Method to Improve the Spatial Resolution of Proton Radiographies

Purpose:We present an optimization method to improve the spatial resolution and the water equivalent thickness accuracy of proton radiographies.Methods:The method is designed for imaging systems measuring only the residual range of protons without relying on tracker detectors to determine the beam trajectory before and after the target. Specifically, the method was used for an imaging set‐up consisting of a stack of 61 parallel‐plate ionization chambers (PPIC) working as a range telescope. The method uses a decomposition approach of the residual range signal measured by the PPIC and constructs subimages with small size pixels geometrically rearranged and appropriately averaged to be merged into a final single radiography. The method was tested using Monte Carlo simulated and experimental proton radiographies of a PMMA step phantom and an anthropomorphic head phantom.Results:For the step phantom, the effective spatial resolution was found to be 4 and 3 times higher than the nominal resolution for the simulated and experimental radiographies, respectively. For the head phantom, a gamma index was calculated to quantify the conformity of the simulated proton radiographies with a digitally reconstructed X‐ray radiography convolved with a Gaussian kernel equal to the proton beam spot‐size. For DTA=2.5 mm and RD=2.5%, the passing ratio was 100%/85% for the optimized/non‐optimized case, respectively. An extension of the method allows reducing the dose given to the patient during radiography acquisition. We show that despite a dose reduction of 25 times (leading to a dose of 0.016 mGy for the current imaging set‐up), the image quality of the optimized radiographies remains fairly unaffected for both the simulated and experimental results.Conclusion:The optimization method leads to a significant increase of the spatial resolution allowing recovering image details that are unresolved in non‐optimized radiographies. These results represent a major step towards clinical application of proton radiography.

Hyperspectral mapping of Iranian east ophiolite mélanges using neutral network classification method

Mapping in the areas, which are near the collision zone, needs extensive and exact studies and tools because of the variety of rocks and complex structures. Hyperspectral sensor is one of the most advanced tools that can do imaging in hundreds of delicate and continuous bands in visible and infrared spectrums. Therefore, it can identify many lithologies despite spectral similarities and complications. In the present study, as one of the first hyperspectral data survey for separating ophiolite melange units in Iran, we applied the spectral-based method of neutral network classification method on hyperion images of Tabas in southeast of Iran. Based on various laboratory–field studies, the lithology of the study area can be separated into five general groups: ophiolite series, metamorphic units, Oligocene–Miocene to Quaternary volcanic units, and lime and flysch units. In order to perform an accuracy assessment of the image processing results, some locations and points were sampled. These samples were analyzed under the microscope and by electron microprobe system. Based on field–laboratory works, the accuracy of the resulting image was validated. Consequently, the neutral network method as advanced hyperspectral image processing method produced an average user accuracy coefficient of 62 % in some parts of colored melanges. Therefore, the method based on neutral network theory for spectral classification has acceptable ability to discriminate various lithologies in the complex geological scenarios.