We present a robust connected-component (CC) based method for automatic detection and segmentation of text in real-scene images. This technique can be applied in robot vision, sign recognition, meeting processing and ...We present a robust connected-component (CC) based method for automatic detection and segmentation of text in real-scene images. This technique can be applied in robot vision, sign recognition, meeting processing and video indexing. First, a Non-Linear Niblack method (NLNiblack) is proposed to decompose the image into candidate CCs. Then, all these CCs are fed into a cascade of classifiers trained by Adaboost algorithm. Each classifier in the cascade responds to one feature of the CC. Proposed here are 12 novel features which are insensitive to noise, scale, text orientation and text language. The classifier cascade allows non-text CCs of the image to be rapidly discarded while more computation is spent on promising text-like CCs. The CCs passing through the cascade are considered as text components and are used to form the segmentation result. A prototype system was built, with experimental results proving the effectiveness and efficiency of the proposed method.展开更多
An integrated novel method of recognizing huge target is described that combines some relatively mature image processing techniques such as edge detection, thresholding, morphology, image segmentation and so forth. Af...An integrated novel method of recognizing huge target is described that combines some relatively mature image processing techniques such as edge detection, thresholding, morphology, image segmentation and so forth. After thresholding the edge image obtained by using Sobel operator, erosion is firstly used to reduce noise and extrusive pixels; then dilation is used to expand some separated pixels into various regions, after that the image segmentation technique is utilized to distinguish the target region with a criterion. The location of the target is also offered. Each technique adopted herein seems not complicated at all, the experimental results demonstrate the efficiency of the combination of these techniques. It is its high computational speed and remarkable robustness resulting from its simplicity that make the method promise to be applied in practical problems requiring real time processing.展开更多
Existing water hazard detection methods usually fail when the features of water surfaces are greatly changed by the surroundings, e.g., by a change in illumination. This paper proposes a novel algorithm to robustly de...Existing water hazard detection methods usually fail when the features of water surfaces are greatly changed by the surroundings, e.g., by a change in illumination. This paper proposes a novel algorithm to robustly detect different kinds of water hazards for autonomous navigation. Our algorithm combines traditional machine learning and image segmentation and uses only digital cameras, which are usually affordable, as the visual sensors. Active learning is used for automatically dealing with problems caused by the selection, labeling and classification of large numbers of training sets. Mean-shift based image segmentation is used to refine the final classification. Our experimental results show that our new algorithm can accurately detect not only ‘common’ water hazards, which usually have the features of both high brightness and low texture, but also ‘special’ water hazards that may have lots of ripples or low brightness.展开更多
文摘We present a robust connected-component (CC) based method for automatic detection and segmentation of text in real-scene images. This technique can be applied in robot vision, sign recognition, meeting processing and video indexing. First, a Non-Linear Niblack method (NLNiblack) is proposed to decompose the image into candidate CCs. Then, all these CCs are fed into a cascade of classifiers trained by Adaboost algorithm. Each classifier in the cascade responds to one feature of the CC. Proposed here are 12 novel features which are insensitive to noise, scale, text orientation and text language. The classifier cascade allows non-text CCs of the image to be rapidly discarded while more computation is spent on promising text-like CCs. The CCs passing through the cascade are considered as text components and are used to form the segmentation result. A prototype system was built, with experimental results proving the effectiveness and efficiency of the proposed method.
文摘An integrated novel method of recognizing huge target is described that combines some relatively mature image processing techniques such as edge detection, thresholding, morphology, image segmentation and so forth. After thresholding the edge image obtained by using Sobel operator, erosion is firstly used to reduce noise and extrusive pixels; then dilation is used to expand some separated pixels into various regions, after that the image segmentation technique is utilized to distinguish the target region with a criterion. The location of the target is also offered. Each technique adopted herein seems not complicated at all, the experimental results demonstrate the efficiency of the combination of these techniques. It is its high computational speed and remarkable robustness resulting from its simplicity that make the method promise to be applied in practical problems requiring real time processing.
基金Project supported by the National Natural Science Foundation of China (Nos. 60505017 and 60534070)the Natural Science Foundation of Zhejiang Province, China (No. 2005C14008)
文摘Existing water hazard detection methods usually fail when the features of water surfaces are greatly changed by the surroundings, e.g., by a change in illumination. This paper proposes a novel algorithm to robustly detect different kinds of water hazards for autonomous navigation. Our algorithm combines traditional machine learning and image segmentation and uses only digital cameras, which are usually affordable, as the visual sensors. Active learning is used for automatically dealing with problems caused by the selection, labeling and classification of large numbers of training sets. Mean-shift based image segmentation is used to refine the final classification. Our experimental results show that our new algorithm can accurately detect not only ‘common’ water hazards, which usually have the features of both high brightness and low texture, but also ‘special’ water hazards that may have lots of ripples or low brightness.
