Wang Weixing

Extracting roads based on Retinex and improved Canny operator with shape criteria in vague and unevenly illuminated aerial images

Abstract. An automatic road extraction method for vague aerial images is proposed in this paper. First, a high-resolution but low-contrast image is enhanced by using a Retinex-based algorithm. Then, the enhanced image is segmented with an improved Canny edge detection operator that can automatically threshold the image into a binary edge image. Subsequently, the linear and curved road segments are regulated by the Hough line transform and extracted based on several thresholds of road size and shapes, in which a number of morphological operators are used such as thinning (skeleton), junction detection, and endpoint detection. In experiments, a number of vague aerial images with bad uniformity are selected for testing. Similarity and discontinuation-based algorithms, such as Otsu thresholding, merge and split, edge detection-based algorithms, and the graph-based algorithm are compared with the new method. The experiment and comparison results show that the studied method can enhance vague, low-contrast, and unevenly illuminated color aerial road images; it can detect most road edges with fewer disturb elements and trace roads with good quality. The method in this study is promising.

Rock fracture image acquisition using two kinds of lighting and fusion on a wavelet transform

In order to obtain more precise information regarding rock fractures, images of a rock sample using two different light sources were acquired and fused. For image acquisition, an epoxy resin liquid was first injected into a fracture zone in situ, and when the epoxy resin was dry, the rock core sample, including the epoxy resin, was removed from the rock base. The rock core sample was then cut into multiple slices and images of the slices using ultraviolet (UV) and visible lighting were acquired. In order to fuse two slice images, an algorithm based on the redundant lifting, non-separable wavelet transform was studied and utilised. Fusion includes three primary steps for each pair of slice images: (1) applying the redundant lifting, non-separable wavelet transform to each image, and then approximating the two images separately; (2) fusing the approximated images corresponding to the decomposition level using certain rules and fusion operators for obtaining fusion coefficients; and (3) applying the fused wavelet coefficients to the redundant lifting non-separable wavelet transform. The results show that combining the proposed method of image acquisition and the image fusion algorithm is not only effective at obtaining a large volume of detailed rock fracture information, it is also an economical and easy to use method. By applying the new method, rock fractures can be easily detected, and many different parameters in different rock engineering applications can be measured and analyzed.

Image enhancement on fractional differential for road traffic and aerial images under bad weather and complicated situations

Abstract To enhance road traffic images and aerial images in bad weather, a new fractional differential operator is studied, and it is different from the traditional Tiansi operator. In the operator, for any sized kernel, the coefficient at the center position is not a constant, and it is the function of the fractional order and the kernel size; for the other pixels in the kernel, there is no zero coefficient value, and the values of the coefficients are calculated according to the distances from the center position on the basis of fractional differential. In experiments, a number of images are tested, and the results show that the studied operator is suitable for the vague images of the low contrast and rich textures, with less color changes. When compared with other newly proposed algorithms (e.g. Dark channel prior and Multiple scale Retinex, etc.) it performs better in some cases.

Remote Sensing Image Automatic Registration on Multi-scale Harris-Laplacian

In order to overcome the difficulty of automatic image registration in image preprocessing, this paper presents an automatic registration algorithm for remote sensing images with different spatial resolutions. The algorithm is studied based on Harris-Laplacian corner detection, which can determine the affine transformation (zoom, rotation, translation) between images of different scales. The corners in the reference and registration images are firstly detected and located by a multi-scale Harris-Laplacian (H-L) corner detector. Secondly, the algorithm chooses SURF (Speeded Up Robust Feature) descriptor to calculate the detected corners descriptors. Then, the multi-resolution corner matching is achieved based on Euclid distance. Finally, according to the LoG (Laplacian Of Gaussian), the scale factor is automatically determined between reference and registration images. A number of remote sensing images are tested, and the experiments show that the studied algorithm can register two remote sensing images of different sizes and resolutions automatically. It also verifies that the algorithm has the lower time cost comparing with the other existing algorithms (e.g. SIFT) within certain detecting accuracy level. This algorithm is also useful for resolving the problem of potential errors due to parallax effects when establishing geometric affine transformation on corners for detecting on buildings with different unknown elevations.