Juichi Miyamichi

Assembly of puzzles using a genetic algorithm

In this paper, we proposed a method for solving the rectangle piece jigsaw puzzle assembly problem. A shape of a piece is a rectangle, and a picture of a puzzle is only painted in black and white, i.e., puzzles are processed as binary images. The assembly of the puzzle is performed only using information of the pixel value on the border line of the pieces. This problem cannot be solved by the simple local piece matching because there are many similar pieces. Global matching is required. The proposed method utilizes a genetic algorithm (GA) to search the optimum piece arrangement because GA has the ability to find the global solution in the large optimization space. The proposed method correctly assembled all pieces in the 8 /spl times/ 8-piece puzzle.

Model-based pose estimation using genetic algorithm

In this paper, we propose a method of the model-based pose estimation in the 3D world from a 2D image. This process consists in searching for the best value of coordinates (x, y, z) and of rotating angles (/spl theta//sub x/, /spl theta//sub y/, /spl theta//sub z/) in which the model object matches most exactly the given input edge image. Taking the match of objects as an index on the six-parameter space, this process can be regarded as a maximum searching problem. We use a genetic algorithm in the estimation of those parameters, and propose a new concept of fitness which takes edge direction into consideration. Our experiments show the results of the proposed method on edge input images.

Assembly of puzzles by connecting between blocks

In this paper, we proposed a method for solving the rectangle piece jigsaw puzzle assembly problem. A shape of a piece is a rectangle, and a puzzle image is RGB full color. The assembly of the puzzle is performed only using information of the pixel value on the border line of the piece. Pieces are connected by a matching function between two pieces. Not only the best matched piece is connected. The matching values to other pieces are used in the evaluation between two pieces. A simple method of puzzle assembly is that a single piece is connected to a block which is defined as a group of connected pieces. But types of piece combinations are restricted in the simple method. In the proposed method, each block is connected each other. Therefore, the best matched connection is selected from many types of combinations between pieces and blocks. The proposed method correctly assembled all pieces in 16 × 12-piece puzzles.

An iterated greedy algorithm for the node placement problem in bidirectional Manhattan street networks

Wavelength Division Multiplexing (WDM) is a technology which multiplexes optical carrier signals on a single optical fiber by using different wavelengths. Lightwave networks based on WDM are promising ones for high-speed communication. If network nodes are equipped with tunable transmitters and receivers, a logical topology can be changed by reassigning wavelengths to tunable transceivers of nodes. Network performance is influenced by the logical node placements. Therefore, an efficient algorithm to obtain the optimal node placement to achieve the best network performance is necessary. In this paper, an iterated greedy algorithm is proposed for this node placement problem. The proposed iterated greedy algorithm consists of two phases, construction and destruction phases. As a local search algorithm, variable depth search is applied after the construction phase. The computational results showed that this iterated greedy algorithm outperformed the best metaheuristic algorithm for this problem.

A new fitness function for shape detection using genetic algorithm

When genetic algorithms (GA) are used for shape detection, the fitness function (evaluation function for chromosomes) must be designed in an appropriate way. Considering shape detection in terms of finding the parameters of model objects to provide the best matching with an input edge image, the chromosome is defined in terms of the objects parameters, and the fitness function is defined in terms of the rate of matching between the edge image of a model object generated using the chromosomes parameters and the input edge image. In this study, the model image and input image are divided into four images each by applying four directional filters, and the fitness function is defined as the sum of the matching rates for each edge direction. Experiments with flat objects were performed to compare the proposed fitness with the conventional technique. The results suggested that the proposed method required slightly more computation time (about 10%) than the conventional fitness function while improving the matching rate from 49% to 93%.

Image mosaicing from a set of images without configuration information

We proposed a method for creating image mosaics automatically under the condition that arrangement of partial images or camera movement is unknown. In this problem, the process can be divided into two steps. The first step is to search rough arrangements of partial images. The next step is to adjust detailed positioning using rough arrangements. In the first step, it is difficult to check up all combinations of arrangements because the number of them is huge. The proposed method utilizes a genetic algorithm (GA) for the first step. The second one for detailed positioning is performed by local search.

Some maximal absolutely unstable sign patterns of 4×4 matrices

In this paper we discuss the absolute instability of the 4×4 qualitative matrices whose diagonal elements are 0, 0, 0 and —. A matrix is said to be stable if all its characteristic values have negative real parts. A sign pattern ( +, −, 0) of a matrix is said to be absolutely unstable if there is no stable matrix which has the sign pattern. Of 531441 different patterns, 299953 are found to be absolutely unstable. By using the overpattern-sub-pattern concept we can define an order relation in the set of qualitative matrices and it is found that there arc 12 basic absolutely unstable sign patterns which are maximal.

Lane Marking Detection by Extracting White Regions with Predefined Width from Bird's-Eye Road Images

Detecting lane markings on roads from in-vehicle camera images is very important because it is one of the fundamental tasks for autonomous running technology and safety driving support system. There are several lane markings detection methods using the width information, but most of these are considered to be insufficient for oblique markings. So, the primary intent of this paper is to propose a detecting lane markings method robust to orientation of markings. In this work, we focus on the width of lane markings standardized by road act in Japan, and propose a method for detecting white lane markings by extracting white regions with constant predefined width from birds-eye road images after segmentation such as categorical color area one. The proposed method is based on the constrained Delaunay triangulation. The proposed method has a merit that can be measure an exact width for oblique markings on the birds-eye images because it can be obtained perpendicular width for edge. The effectiveness of the proposed method was shown by experimental results for 187 actual road images taken from an in-vehicle camera.

Run-based volume intersection for shape recovery of objects from their silhouettes

Volume intersection (VI) is a successful technique for reconstructing 3-D shapes from 2-D images (silhouettes) of multiple views. It consists of intersecting the cones formed by back-projecting each silhouette. The 3-D shapes reconstructed by VI are called visual hull (VH). In this paper we propose a fast method obtaining the VH. The method attempts to reduce the computational cost by using a run representation for 3-D objects called SPXY table that is previously proposed by us. It makes cones by back-projecting the 2-D silhouettes to the 3-D space through the centers of the lens and intersects them keeping the run representation. To intersect the cones of multiple views keeping the run representation, we must align the direction of runs representing the cones. To align them we use the method of swapping two axes of a run-represented object at the time cost of O(n) where n is a number of runs, which is also previously proposed by us. The results of experiments using VRML objects such as human bodies show that the proposed method can reconstruct a 3-D object in less than 0.17 s at the resolution of 220 × 220 × 220 voxels from a set of silhouettes of 8 viewpoints on a single CPU.

Construction of neural networks to select harmonious color combinations

Humans have a sense of color which, when a color is presented, can determine another color that is harmonious to the presented color (called harmonious color combination). This paper considers the two-color combination problem, which is to decide whether or not the two given colors are harmonious. The three-layered neural network with the error backpropagation learning is used to realize on a computer the color combination aesthetics of an individual trained in color harmonization. The performance of the realized system is examined. The result obtained by the constructed system is the correct recognition rate of 97 percent for the learned data and approximately 70 percent for the unlearned data. The correct recognition rate for the unlearned data agrees almost with the rate of coincidence when the same subject selected the harmonious color twice. It is observed also that the color selection histogram pattern of the neural network is very similar to that of the subject. It is possible to identify the original subject from the tendency of the color selection tendency of the neural network. Thus, it is verified that the model is constructed reflecting well the color selection aesthetics of the subject.