Object extraction is an important tool in many applications within the image processing and computer vision communities. You Only Look Once version 3 (YOLOv3) has been extensively applied to many fields as a state-of-the-art technique for object semantic detection. Despite its numerous characteristics, YOLOv3 has to be combined with appropriate image segmentation technologies to achieve effective 2D object extraction in real-time monitoring, robot navigation, and target search. In this paper, the K-means algorithm is applied to the segmentation of depth images. Considering the inherent sensitivity to the randomness of the initial cluster center and the uncertainty of cluster number K in the initialization phase of the K-means algorithm, this paper proposes a new method that combines the semantic image information with the image depth information. Specifically, this method proposed to pre-classify the center depth of the object to determine the appropriate value of K required in the K-means algorithm. At the same time, the proposed algorithm improves the selection of the initial center via the maximin method. This paper introduces a multi-parameter extraction method to enable to correctly identify the object of interest after image segmentation. The technique considers three parameters to achieve this: i) the elements of size, ii) the connected domain, and iii) the diagonal detection. Experiments using open-source datasets demonstrate that the average processing time and the segmentation accuracy of the improved K-means algorithm are 20.36% faster and 3.12% higher than the conventional K-means algorithm, respectively. The extraction accuracy of the proposed method is 6.69% higher than that of the SuperCut extraction method.

Visual Simultaneous Localization and Mapping (SLAM) technologies based on point features achieve high positioning accuracy and complete map construction. However, despite their time efficiency and accuracy, such SLAM systems are prone to instability and even failure in poor texture environments. In this paper, line features are integrated with point features to enhance the robustness and reliability of stereo SLAM systems in poor texture environments. Firstly, method Edge Drawing lines (EDlines) is applied to reduce the line feature detection time. Meanwhile, the proposed method improves the reliability of features by eliminating outliers of line features based on the entropy scale and geometric constraints. Furthermore, this paper proposes a novel Bags of Word (BoW) model combining the point and line features to improve the accuracy and robustness of loop detection used in SLAM. The proposed PL-BoW technique achieves this by taking into account the co-occurrence information and spatial proximity of visual words. Experiments using the KITTI and EuRoC datasets demonstrate that the proposed stereo Point and EDlines SLAM (PEL-SLAM) achieves high accuracy consistently, including in challenging environments difficult to sense accurately. The processing time of the proposed method is reduced by 9.9% and 4.5% when compared to the Point and Line SLAM (PL-SLAM) and Point and stereo Point and Line based Visual Odometry (sPLVO) methods, respectively.

To solve the self-alignment problem of the Strapdown Inertial Navigation System (SINS), a novel adaptive filter based on Complementary Ensemble Empirical Mode Decomposition (CEEMD) is proposed. The Gravitational Apparent Motion (GAM) is used in the coarse alignment, and the problem of obtaining the attitude matrix between the body frame and the navigation frame is attributed to obtaining the matrix between the initial body frame and the current navigation frame using two gravitational apparent motion vectors at different moments. However, the accuracy and time of this alignment method always suffer from the measurement noise of sensors. Thus, a novel adaptive filter based on CEEMD using an $l_2$ -norm to calculate the similarity measure between the Probability Density Function (PDF) of each Intrinsic Mode Function (IMF) and the original signal is proposed to denoise the measurements of the accelerometer. Furthermore, the advantage of this filter is verified by comparing with other conventional denoising methods, such as PDF-based EMD (EMD-PDF) and the Finite Impulse Response (FIR) digital low-pass filter method. The results of the simulation and experiments indicate that the proposed method performs better than the conventional methods in both alignment time and alignment accuracy.

Vehicles have to rely on satellite navigation in an open environment. However, satellite navigation cannot obtain accurate positioning information for vehicles in the interior of underground parking lots, as they comprise a semi-enclosed navigation space. Therefore, vehicular navigation needs to take into consideration both outdoor and indoor environments. Actually, outdoor navigation and indoor navigation require different positioning methods, and it is of great importance to choose a reasonable navigation and positioning algorithm solution for vehicles. Fortunately, the integrated navigation of the Global Positioning System (GPS) and the Micro-Electro-Mechanical System (MEMS) inertial navigation system could solve the problem of switching navigation algorithms in the entrance and exit of underground parking lots. This paper proposes a low cost vehicular seamless navigation technology based on the reduced inertial sensor system (RISS)/GPS between the outdoors and an underground garage. Specifically, the enhanced RISS is a positioning algorithm based on three inertial sensors and one odometer, which could achieve a similar location effect as the full model integrated navigation, reduce the costs greatly, and improve the efficiency of each sensor.