Multi-Lingual Character Recognition and Extraction using Recurrent Neural Networks
A Novel Approach to Smart Autonomous Monitoring of Indoor Plant Health Based on Leaf Color
Edge Preserving Image Denoising with Deep Networks and Ensemble Convolutional Neural Network Architectures
Neural Network Approach to Detect Renal Calculi
Enhancing Fabric Quality Control: Implementing Real-Time Defect Detection with Image Processing Techniques and Arduino
Identification of Volcano Hotspots by using Resilient Back Propagation (RBP) Algorithm Via Satellite Images
Data Hiding in Encrypted Compressed Videos for Privacy Information Protection
Improved Video Watermarking using Discrete Cosine Transform
Contrast Enhancement based Brain Tumour MRI Image Segmentation and Detection with Low Power Consumption
Denoising of Images by Wavelets and Contourlets using Bi-Shrink Filter
Contour representation of binary object is widely used in pattern recognition. Chain codes are compression methods, where the original data are reconstructed from the compressed data for representing binary objects including contours. Though a notably huge image size reduction is obtained by fixed-length chain code, so far, more efficient and reliable methods for data encoding is possible by using technique that treats the binary bits differently considering its requirement of storage space, energy consumption, speed of execution, and so on. This paper proposes a new variant of Huffman Coding (HC) by taking into consideration the fact that the costs of bits are different, the new representation of the Freeman Chain Code (FCC) is based on an eight-direction scheme. An experimentation of the cost efficiency of the new representation over the classical FCC is described and compared to other techniques. The experiments yield that the proposed FCC representation reduces overall both the storage and the transmission cost of encoded data considerably compared to the classical FCC.
The efficiency of a palmprint recognition system lies in its robustness and antispoofing capabilities. Enhancing their capabilities will require the use of new techniques for extracting more discriminative features from collected images particularly those with different illuminations. Feature level fusion is presented using the popular Gabor Wavelet Transform (GWT) for palmprint images collected from different light illumination, such as Red, Green, Blue, and Near Infra-Red (NIR). Individual spectra were fused as triple (R, B, NIR, and G, B, NIR) at feature level in the experiments, followed by verification of system performance by a number of classifiers. The method demonstrates that an increase in recognition performance of almost 100% could be obtained in the system by fusion of the spectra when compared to previous works.
The breast cancer is absolutely life intimidating and dreadful disease. The primary screening of breast tumor is still under research because of some risk features, such as gene, taking birth control pills, smoking, obesity, and age are playing vital role spreading the cancers. The malignant tumors induct into the breast cells and eventually this tumor extends to the surrounding tissues. The proposed technique consists of four steps. Step 1 is for digitized noises removal, step 2 is for suppression of radio opaque artifacts, step 3 is for Pectoral Muscle removal, and step 4 is for detecting location of cancer on breast for emphasizing the region of breast profile. To reveal the capability of this technique, two separate digital mammograms are tested using GT (Ground Truth) mammograms for assessment of performance characteristics. The Experimental results indicate that the breast cancer regions are extracted truthfully in compliance to respective Ground Truth Images.
Capturing the image in fog scene suffers from the distortions of the information and the time taken to predict the object in the way become complicated. To overcome this difficulty, the images are taken into consideration as new a technology opens door for converting the analog scene to digital scene in the form of an image. The image captured in less visibility of the scene predominantly in foggy weather conditions, the structure of image, and also several human activities like drones, aircrafts, flights, and travellers affect several computer vision applications like tracking, artificial intelligence, remote sensing, etc. Thus restoring back the outdoor scene from such foggy image is significantly important. The main focus is to defog the image in the patch; atmospheric light in foggy days looks as that of the fog, so this to be reduced; edges and corners must be visible, inner depths to be reconstructed from the fog scene. To fulfill these characteristics of image, several defog techniques were investigated. The Spatial Random Markov Fields with Boundary Constraints was proposed, which performs on image within the space and patch using boundary constraints. Experimental results demonstrate that the proposed work is efficient to remove fog, restore space of an image, and preserve the natural atmospheric light even in foggy days without changing the color.
Land use/land cover (LULC) information in the form of maps is essential for the planning, supervising natural resources, utilisation of land to meet the increasing human demands and monitoring changes in the ecosystem. In this study, remote sensing data and geographic information system applications were used to analyse the LULC and its changes in Tirupati, which is located in the Seshachalam hill range in the Chittoor district of Andhra Pradesh (AP) state, South India. The study area is a world-famous pilgrimage centre and fast-developing town. Therefore, updated LULC maps must be created for various departments. The aim of this study was to classify and determine changes in the LULC over the 40-year period 1978-2018 by using multi-temporal Landsat satellite images and Survey of India toposheet map. The 1978 and 2018 Landsat images and field survey data were selected to classify the data. The ERDAS Imagine v16 and ArcGIS v10.1 were used to process images and assess the changes in land use of this study area. Classification was performed using the maximum likelihood classifier algorithm of supervised classification. Images were classified into five major classes: forest, water bodies, agricultural land, barren land, and built-up land. A post-classification change detection technique was used here to find changes in LULC. Changes were mainly observed in the built-up areas. The results demonstrate that during the forty years period, built-up area and barren land/other land increased 454.33%, and 104.7%, and area under waterbodies, agriculture, and forest decreased to 73.07%, 61.84%, and 31%, respectively. In future, these changes may have a significant influence on the ecosystem.
The fast growth in internet and communication technology has facilitated an increase in the exchange of digital multimedia content like audio, video, images, etc. It is necessary to accomplish secure communication for digital information in open networks. The art of hiding information has become a major issue in the late years, the security of information has become a big interest in this internet era. Encoding information using QR codes were done, as they have greater potential to carry different types of information in a smaller space. Due to these properties, they have gained popularity in various fields of application and are used for both human interaction and automated systems. This paper includes two phases, i.e.; generating a QR code and hiding the QR code inside a color image. In the embedding process, binary image is converted into a corresponding digitally invisible watermark that is inserted in a QR code. The Quick response (QR) code (watermark) is embedded by using the Discrete Wavelet Transform in YCbCr color space, which should be extracted blindly without the host image or original watermark after applying different image processing attacks.
