Search

In recent years, due to frequent accidents in mine, it is significantly meaningful to develop monitoring algorithms that can improve safeties. However, the video signal in mine is usually strongly polluted during its sampling and transmission, resulting in difficulties for monitoring. This study presents a mine safety monitoring algorithm. Our approach has two parts, i.e., preprocessing and recognition of visual information. First, we address a video enhancement method that uses inter-frame similarity prediction to accelerate non-local means, second, we propose an object recognition method based on improved Hough forest, it speeds up recognition process by reducing the size of search window, and it increases the recognition accuracy through introducing time-dimensional analysis. Experimental results illustrate that the proposed algorithm obtains 66% reduction in denoising time and an improved recognition rate. It is reasonable and reliable to apply our algorithm to mine safety monitoring.

In this paper, we present a novel approach to motion estimation (ME) of arbitrarily shaped video objects in wavelet domain. We explore the guiding role of binary alpha-plane in assisting ME of video objects and first devise a new block matching scheme of alpha-plane, by exploiting boundary expansion and boundary masks. To eliminate shift-variance, we modify low-band-shift (LBS) method via substituting variable-size block for wavelet block. Combining the modified LBS with a hierarchical structure, we further present a multiscale ME approach. Extensive experiments show that the proposed approach outperforms most of previous methods in terms of both subjective quality and objective quality. Moreover, significant reduction is achieved in computational complexity (89.05% at most) and memory requirement.

This paper proposes an accuracy-distortion model for overcomplete wavelet-based scalable motion estimation by exploiting the theory of stationary random process. We first estimate the motion compensation errors in spatial domain due to inaccurate motion vectors, then extend the results to overcomplete wavelet domain, and further derive the errors caused by fraction-pixel motion vectors. Finally, combining with a visibility model of wavelet coefficient errors, this paper proposes a novel algorithm to estimate the motion vector accuracy with which the motion compensation errors will be invisible. Experimental results show that the proposed algorithm is effective in estimating the visually lossless accuracy threshold of motion vectors. The proposed algorithm can be used in scalable coding of motion vectors. Also it can accelerate the motion estimation speed by stopping halfway at the accuracy that will not cause any visible errors.

In this paper, we propose a fast motion estimation algorithm of arbitrarily shaped video object in MPEG-4. The proposed algorithm incorporates the binary alpha-plane and the extended contour to predict accurately the motion vectors of boundary macroblocks so that the conventional fast motion estimation algorithms can be employed to search the motion vectors of opaque macro-blocks using the motion vectors of the neighboring boundary macro-blocks as the initial center. Experimental results show that the proposed algorithm requires low computation complexity while provides good motion compensation quality