41 research outputs found
Accuracy of Various Methods to Estimate Volume and Weight of Symmetrical and Non-Symmetrical Fruits using Computer Vision
Get PDFMany researchers have used images to measure the volume and weight of fruits so that the measurement can be done remotely and non-contact. There are various methods for fruit volume estimation based on images, i.e., Basic Shape, Solid of Revolution, Conical Frustum, and Regression. The weight estimation generally uses Regression. This study analyzed the accuracy of these methods. Tests were done by taking images of symmetrical fruits (represented by tangerines) and non-symmetrical fruits (represented by strawberries). The images were processed using segmentation in saturation color space to get binary images. The Regression method used Diameter, Projection Area, and Perimeter as features that were extracted from the binary images. For symmetrical fruits, the best accuracy was obtained with the Linear Regression based on Diameter (LDD), which gave the highest R2 (0.96 for volume and 0.93 for weight) and the lowest RMSE (5.7 mm3 for volume and 5.3 gram for volume). For non-symmetrical fruits, the highest accuracy for non-symmetric fruits was given by the Linear Regression based on Diameter (LRD) and Linear Regression based on Area (LRA) with an R2 of 0.8 for volume and weight. The RMSE for LRD and LRA for strawberries was 3.3 mm3 for volume and 1.4 grams for weight
THICKNESS ASSESSMENT OF SKIN LESION USING 3D SKIN SURFACE IMAGING
No full textThickness is one of the parameter to be assessed to determine the severity level
and treatment efficacy of skin lesion. In practice, the dermatologists subjectively
assess the skin lesion thickness using the tactile inspection which does not outcome a
quantitative measure thus result inter-rater and intra-rater variation. The challenge in
measuring thickness of skin lesions is that the surface curvature of human skin is
undulate in 3D and varying at different body location. Several researches to
objectively measure the skin lesion thickness has been developed by imaging the skin
structure using microscope, X-Ray and ultrasound. However, the researches are not
adapting the tactile inspection. In this research, an objective method to measure skin
lesion thickness and determine the range of lesion thickness for each severity level on
the standard Psoriasis Area and Severity Index (PASI) assessment tool by adapting
the dermatologists’ tactile inspection is developed. The method obtains the 3D
surface profile of skin lesion by using a 3D fringe projection scanner. The thickness
of skin lesion is determined by averaging the elevations of lesion’s surface from an
estimated lesion base (the assumed skin surface without the lesion)
Face Detection of Thermal Images in Various Standing Body-Pose using Facial Geometry
Get PDF Automatic face detection in frontal view for thermal images is a primary task in a health system e.g. febrile identification or security system e.g. intruder recognition. In a daily state, the scanned person does not always stay in frontal face view. This paper develops an algorithm to identify a frontal face in various standing body-pose. The algorithm used an image processing method where first it segmented face based on human skin’s temperature. Some exposed non-face body parts could also get included in the segmentation result, hence discriminant features of a face were applied. The shape features were based on the characteristic of a frontal face, which are: (1) Size of a face, (2) facial Golden Ratio, and (3) Shape of a face is oval. The algorithm was tested on various standing body-pose that rotate 360° towards 2 meters and 4 meters camera-to-object distance. The accuracy of the algorithm on face detection in a manageable environment is 95.8%. It detected face whether the person was wearing glasses or not
Deteksi Gulma Berdasarkan Warna HSV dan Fitur Bentuk Menggunakan Jaringan Syaraf Tiruan
Get PDFGulma merupakan tanaman pengganggu dalam lahan pertanian. Herbisida merupakan obat yang efektif membunuh gulma tersebut. Penyemprotan herbisida harus tepat sasaran kepada gulma saja dan tidak mengenai tanaman. Penelitian ini membuat sistem yang dapat mendeteksi gulma secara otomatis di antara tanaman pada lahan pertanian riil. Sistem ini menggunakan gambar lahan pertanian riil dimana tanaman tampak utuh (daun dapat lebih dari satu) yang diambil menggunakan kamera dengan posisi vertikal menghadap ke bawah. Algoritma yang dibuat menggunakan segmentasi berdasarkan warna hijau dalam ruang warna HSV untuk mendeteksi daun, baik gulma maupun tanaman pada beragam pencahayaan. Sebanyak tiga fitur bentuk domain spasial digunakan untuk membedakan gulma dengan tanaman yang memiliki karakteristik bentuk daun yang berbeda. Fitur bentuk yang digunakan adalah Rectangularity, Edge-to-Center distances function, dan Distance Transform function. Klasifikasi gulma dan tanaman menggunakan metode Jaringan syaraf tiruan (JST) yang dapat dilatih secara offline. Dari 149 tanaman yang terdeteksi dimana 70% sebagai data training, 15% data validasi dan 15% data uji, didapati akurasi pengujian sebesar 95.46%.AbstractWeed is a major challenge in a crop plantation. A herbicide is the most effective substance to kill this unwanted vegetation. Spraying the herbicide must be done carefully to target the weeds only. Here in this research, we develop an algorithm that detects weeds among the plants based on the shape of their leaves. The detection is based on images that were acquired using a camera. The leaves of weeds and plants were detected based on their green color using segmentation in HSV color-space as it is more effective to detect objects in various illumination. Three shape features were extracted, which are Rectangularity that is based on Rectangularity, Edge-to-Center distance function, and Distance Transform function. Those features were fed into a learning algorithm, Artificial Neural Network (ANN), to classify whether it is the plant or the weed. The testing on the weed classification in a real outdoor environment showed 95.46% accuracy using a total of 149 detected plants (70% as training data, 15% as validation data, and 15% as testing data)
Automatic Estimation of Human Weight From Body Silhouette Using Multiple Linear Regression
No full textClassification of lung condition for early diagnosis of pneumonia and tuberculosis based on embedded system
Get PDFThe lungs are the main organs in the respiratory system that have a function as a place for exchange of oxygen and carbon dioxide. Due to the importance of lung function, indications of lung disorders must be detected and diagnosed early. Research on the classification of lung conditions generally uses chest x-ray image data. Where a time-consuming procedure is needed to obtain the data. In this research, an embedded system to diagnose lung conditions was designed. The system was made to be easy to use independently and provides real-time examination results. This system uses parameters of body temperature, oxygen saturation, fingernail color and lung volume in classifying lung conditions. There are three conditions that can be classified by the system, that is healthy lungs, pneumonia and tuberculosis. The k-nearest neighbor method was used in the classification process in the designed system. The dataset used was 51 data obtained from the hospital. Each data already has a label in the form of lung condition based on the doctor’s diagnosis. The proposed system has an accuracy of 88.24% in classifying lung conditions
Deteksi Kesegaran Ikan Tongkol (Euthynnus Affinis) secara Otomatis Berdasarkan Citra Mata Menggunakan Binary Similarity
Get PDFIkan tongkol (Euthynnus Affinis) adalah salah satu ikan yang paling banyak diminati di Indonesia karena kandungan proteinnya hampir setara ikan tuna, namun dengan harga relatif lebih murah. Ikan termasuk komoditi pangan yang mudah rusak tanpa adanya penanganan khusus ketika ikan ditangkap. Padahal, mutu dan nilai jual ikan sangat tergantung dari parameter kesegaran ikan itu sendiri. Penelitian ini mengembangkan metode deteksi kesegaran ikan tongkol menggunakan fitur berupa citra mata ikan. Mata ikan dapat digunakan untuk mengetahui tingkat kesegarannya. Ikan segar memiliki pupil bulat berwarna hitam yang utuh dan jernih di tengahnya. Hal tersebut kemudian dijadikan knowledge-based dari proses deteksi kesegaran ikan. Sebelum dilakukan proses deteksi, dilakukan proses pre-processing untuk mendapatkan gambar kepala ikan secara otomatis. Selanjutnya dilakukan perhitungan similarity antara citra biner kepala ikan dengan 2 buah template, yakni Template-Mata untuk mendeteksi mata dan Template-Tengah untuk mendeteksi bulat hitam di tengah mata. Sebanyak 30 citra mata ikan dengan kriteria segar dan tidak segar digunakan sebagai data pengujian. Dari pengujian, kedua template tersebut mampu membedakan ciri morfologis dari mata ikan yang segar dengan tepat.AbstractTongkol fish (Euthynnus Affinis) is one of the most popular fish in Indonesia because it has more protein than tuna, but with a relatively cheaper price. Fish is a perishable food commodities if it is caught without any special handling. In fact, the quality and value of fish selling depends on the parameters of the freshness of the fish itself. This study developed a method for detecting freshness of tongkol fish using features that is extracted from the image of a fish\u27s eye. Fish eye can be used to determine the level of freshness. Fresh fish have whole round and clear black pupils in the middle. This is then made into knowledge-base on the process of detecting the freshness. First, this fully automatic detection performed a pre-processing process to obtain automatic fish head images. It was then compared with two templates, which are eye-template and middle-template. If the fish head image has similarity below certain threshold then it is classified as fresh fish, or else it is non-fresh fish. A total of 30 images of fish with fresh and non-fresh criteria were used as test data. From the test, the two templates can classify the morphological characteristics of fresh fish eyes precisely
Sistem Deteksi Jumlah, Jenis dan Kecepatan Kendaraan Menggunakan Analisa Blob Berbasis Raspberry Pi
Get PDFPenghitungan kondisi lalu lintas guna analisa kualitas jalan raya umumnya dilakukan secara manual. Hal ini tentunya membutuhkan biaya dan SDM yang tinggi serta tidak dapat dianalisa secara langsung. Dalam penelitian ini telah dikembangkan metode pengenalan jenis, jumlah dan kecepatan kendaraan secara otomatis menggunakan pengolahan citra digital. Metode yang berdasarkan analisa terhadap BLOB (Binary Large OBject) tersebut ditanamkan pada sistem berbasis Raspberry Pi. Setiap blob merupakan connected-component yang diperoleh dari proses thresholding terhadap perubahan nilai pixel dari sebuah frame dan frame rujukan dalam metode background subtraction. Jenis kendaraan ditentukan oleh jumlah piksel dalam bounding-box setiap blob. Jumlah kendaraan yang melaju dihitung dengan memberikan garis virtual dimana jumlahnya akan bertambah jika centroid dari setiap bounding-box kendaraan melewatinya. Kecepatan kendaraan dihitung dengan membagi jarak sebenarnya dari koordinat awal hingga garis virtual sepanjang 12 meter yang dibagi dengan waktu centroid tersebut untuk menempuhnya. Algoritma tersebut diimplementasikan pada sistem berbasis Raspberry Pi dengan input kamera yang terhubung dengan serial monitor untuk menampilkan output penghitungan. Pengujian akurasi deteksi jenis kendaraan yakni sepeda motor, kendaraan ringan dan berat menghasilkan akurasi 93,39%. Pengujian jumlah kendaraan menghasilkan rata-rata akurasi 93,48% untuk semua jenis kendaraan. Pengujian laju kendaraan yang dideteksi dengan dibandingkan kecepatan pada spedometer kendaraan menunjukkan akurasi 93,9%. AbstractAn analysis on traffic condition usually carried out manually by visual observation. This method demands high human resource and cannot be analysed immediately. This paper present an algorithm to analyse type, number and speed of vehicles that passing by a road automatically using BLOB (Binary Large Object) analysis. Each blob is a connected-component as a result of thresholding after background subtration process. Type of vehicles was determined by measuring pixel number of blob’s bounding box. Number of vehicles was determined by drawing virtual line where the number was increased once a centroid of bounding box passed it. Speed of vehicles was determined using basic speed formula where 12 meters of actual distance between the beginning coordinate and virtual line was divided by time to travel it. The algorithm was embedded in Raspberry Pi where videos were acquired using attached web camera. The analysis result was shown in connected serial monitor. Testing on vehicles’ type detection (motorcycle, light vehicle, heavy vehicle) result accuracy of 93.9%, number of vehicles result accuracy of 93.48%, whilst speed of vehicles result accuracy of 93.9%
Sistem Object Tracking pada Quadcopter Menggunakan Segmentasi Citra dengan Deteksi Warna HSV dan Metode Regresi Linier Berbasis Raspberry Pi
Get PDFSaat ini, banyak aplikasi perangkat cerdas yang dikembangkan untuk melakukan tugas secara mandiri tanpa menerima perintah dari manusia. Oleh karena itu, mengembangkan sistem yang memungkinkan perangkat untuk melakukan tugas pengawasan seperti mendeteksi dan melacak objek bergerak akan memungkinkan tugas yang lebih canggih untuk diterapkan pada robot di masa depan. Teknologi Quadcopter sesungguhnya dapat memudahkan pekerjaan manusia dalam melakukan pengawasan dan pelacakan seperti pada kasus pelacakan lansia atau ABK (Anak Berkebutuhan Khusus) secara otomatis agar kerabat dapat melakukan pengawasan dengan menggunakan drone. Sehingga penelitian ini dilakukan untuk membuat sebuah sistem pada drone atau quadcopter agar dapat mendeteksi objek dan mengikutinya. Pada implementasinya, orang yang berkebutuhan khusus dan membutuhkan pengawasan akan mengenakan atribut berupa topi dengan warna solid. Warna topi tersebut akan dijadikan acuan untuk threshold segmentasi warna untuk mendeteksi objek topi tersebut dengan pemrosesan citra digital. Pergerakan drone ditentukan oleh prediksi jarak, sudut, dan ketinggian objek berdasarkan regresi linier yang dihasilkan dari 123 data latih. Hasil deteksi sistem juga cukup sesuai dengan pergerakan drone ketika diuji dengan 27 data. Akurasi dari prediksi gerak pitch adalah 84%, prediksi gerak yaw adalah 94%, dan prediksi gerak up/down adalah 91,5%. Adapun waktu komputasinya adalah 0.175829662 detik per frame. Abstract Nowadays, many intelligent device applications are developed to perform tasks independently without receiving commands from humans. Therefore, developing systems that allow devices to perform surveillance tasks such as detecting and tracking moving objects will allow more sophisticated tasks to be applied to robots in the future. Quadcopter technology can actually facilitate human work in monitoring and tracking, such as in the case of tracking the elderly or children with special needs automatically so that relatives can carry out surveillance using drones. So this research was planned to create a system on a drone so it can detect objects and follow them. In its implementation, people with special needs and need supervision will wear an attribute in the form of a hat with a solid color. The color of the hat will be used as references for the color segmentation threshold to detect the hat object with digital image processing. The movement of the drone is determined by the prediction of the distance, angle, and height of the object based on linear regression generated from 123 training data. The system detection results are also quite in accordance with the movement of the drone when tested with 27 data. The accuracy of pitch motion prediction is 84%, yaw motion prediction is 94%, and up/down motion prediction is 91.5%. The computation time is 0.175829662 seconds per frame
