Most force sensors that have been developed include electric devices such as strain gauges. However, electrical devices are not always suitable for medical devices. Electrical force sensors need amplifiers and wiring for signal transfer. This causes the size and cost of the overall system to become large. In addition, sterilizing or disinfecting such electrical devices is difficult. In special environments such as those of magnetic resonance imaging (MRI), the use of electricity should be avoided. One solution to these issues is to not use electrical parts such as wiring and circuits. Takaki et al. [6] developed a force sensor based on force visualization by using moir�� fringe patterns. Tadano and Kawashima [7] developed a system to generate feedback on force sensation without a force sensor by utilizing a pneumatic servo system.

Kawahara et al. [8] developed a system to measure the stiffness of an organ by pushing the organ using air and capturing the deformation using a camera. Peirs et al. [9] developed a force sensor that detects the deformation of a flexible structure by using optical fibers. Using fibers leads to issues pertaining to wiring, such as signal distortion resulting from bending, twisting, and chirping. Tada et al. [10] developed a force sensor that functions in MRI environments. A point light source is attached to tip of an elastic frame. If a force is applied to the elastic frame, the position of peak illumination changes. By detecting the change via photosensors, the applied forces can be estimated.

These sensors are mainly used for laparoscopic surgery, and the size of the parts and the range of measurable forces are different. The part sizes and forces measured for laparoscopic surgery are larger (in the cm and N ranges, respectively) than the corresponding values that would be ideal for endoscopy in neurosurgery (in the mm and mN ranges) Sensors utilizing visualization of force have been developed, although the purpose of such sensors is not medical. Ohka et al. [11] have developed a three-axis force sensor by observing the states of conical feelers using a camera. Kamiyama et al. [12] have developed a sensor that can measure the direction, magnitude, and distribution of force by observing two layers of spherical markers using a camera.

However, to apply these concepts to force sensors in small and thin fiberscopes, the sensor size GSK-3 should be reduced and a method to construct small markers must be developed. In general, this issue is considered to be the disadvantage of sensors utilizing force visualization, as mentioned in previous studies [5,13]. On the other hand, we previously developed a robotic system with force sensor and feedback systems for neurosurgery [14�C16]. Unfortunately, the developed force sensor was based on a strain gauge system, and as a result, the above issues of sterilization and MRI compatibility were not resolved.

Growing cities also have a desire to control development near greenbelt areas [2]. Tree species maps can also be used by conservationists hoping to protect the favored nesting place of a particular species of bird [3]. Thus, there is demand for accurate and up-to-date land cover maps. Remote sensing approaches have proven to be valuable in developing land cover maps compared to traditional methods [5, 4]There is a considerable amount of literature regarding the identification and classification of tree species utilizing airborne or space-borne imagery using numerous classification methods. Generally, tree species identification using remote sensing data depends upon spatial, spectral and temporal resolution.

In addition, several authors discuss the importance of different classification algorithms and supplementary data such as LiDAR for the identification of tree species.The first major use of digital imagery and machine processing was to map vegetation health a year after the corn leaf blight in 1970 [6]. The launch of Landsat in 1972 began a serious investigation into the capabilities of remote sensing for vegetation management. In 1978, Kan and Weber released their study on mapping forests and rangelands using Landsat. They found they could separate hardwood forest, softwood forest and grasslands with 70% accuracy [7].Meyer, Staenz, and Itten used color-infrared film to image two areas of the Swiss Plateau and were able to classify 5 classes of trees with 80% accuracy [8].

In another study, Cypress and Tupelo trees were mapped utilizing moderate spatial resolution Landsat TM imagery in an effort to develop a method of locating wetland areas for more effective land management [9]. Higher spatial resolution AV-951 imageries were also used by several authors for tree species identification [11, 10]. Carleer and Wolff attempted an analysis of tree species in a Belgian forest using a high-resolution IKONOS image [12]. They suggest that forest tree mapping requires higher spatial and spectral resolution.Combinations of different date and spatial resolution multi-spectral images have been used for species classification [1].

They found that the images taken in September were most useful in identification of tree species and 1m spatial resolution is optimal for reducing the shadow effects in between the trees in Columbia, Missouri. Fall imagery appeared to provide the most Dacomitinib information for species identification while spring leaf-out imagery was next best in terms of species identification [13].Spectral resolution is also a significant factor in determining overall classification accuracy. Comparatively few authors have used hyperspectral imagery for tree species identification.