Search

Video cameras are among the most commonly used devices throughout the world which results in imaging technology being one of the most important areas for research and development. Imaging technology requires constant research as it is used in crucial applications such as video conferencing and surveillance. In the field of image processing, motion detection and estimation are fundamental steps in extracting information on objects segmented from their backgrounds. In this paper, a cohesive approach is presented that uses two algorithms for motion estimation and detection. The proposed method is able to detect moving objects using maximum average correlation height (MACH) filter. Upon obtaining the accurate coordinates of an object of interest from the MACH filter, the next part of the algorithm starts tracking the object. For tracking, a particle filter is used to estimate the motion of the object using a Markov chain. To enhance the accuracy of particle filter, an approximate proximal gradient algorithm is employed for unconstrained minimization of the particles which restricts the tracking process to target templates (most essential information) only. Finally, a comparison between the proposed algorithm and recent similar algorithms is made that demonstrates the minimization of tracking errors using the proposed technique. [ABSTRACT FROM AUTHOR]

The image transformation with respect to mammalian retina-based sensor geometry enables us to resolve some major difficulties in object recognition. Logarithmic mapping is a mathematical transform that converts scale and angular variations to image shifts which helps in classifying target objects correctly. The primary issues dealt with in the design of correlation filters include the ability to suppress clutter and noise, easy detection of the correlation peaks and distortion tolerance. The correlation pattern recognition concept is applied for developing filters that provide invariance to distortions by adopting a suitable training mechanism. But, a separate training requirement for each distortion type is a drawback in correlation filters. By joining a recent variant of correlation filters, i.e. the Extended Maximum Average Correlation Height Filter with the strengths of logarithmic translation, we get a supplementary advantage which is useful in most of the recognition applications. The proposed combination yields a design which can manage multiple distortions in a single training instance with pronounced detection results against complex input scenes. [ABSTRACT FROM AUTHOR]

The automatic classification of bone texture into healthy or osteoporotic cases presents a major challenge since there is no visual difference between the two cases. This classification requires an inspection of the fine granularity in the bone radiographs which is usually difficult with a naked eye. We have proposed a novel method in this paper, that can be used for the classification of bone radiographs into healthy or osteoporotic cases. We mimic the observations of the physicians by preprocessing the bone radiographs with Gabor filters bearing a high frequency. Later, we design and utilize a convolutional neural network wherein filtered images are fed as input to the system which classifies the images into their respective classes. The proposed algorithm has been validated on a bone radiograph challenge dataset. Our results depict that the method proposed in this research exhibits very good results in terms of classification. A comparison of the proposed and the contemporary research methods has also been shown in this paper. The experimental results show that by exploiting high frequency Gabor filters and employing the convolutional neural network architecture, good results in performing the classification of bone radiographs are achieved. [ABSTRACT FROM AUTHOR]