Elsevier

Neurocomputing

Volumes 52–54, June 2003, Pages 561-566
Neurocomputing

A reconstruction method of projection image on worker honeybees’ compound eye

https://doi.org/10.1016/S0925-2312(02)00785-3Get rights and content

Abstract

Visual perception by insect compound eyes is of interest in visual science for revealing neuronal principals of vision, and applying their small and simple structures to artificial vision systems. In order to investigate how animal behavior is related to sensory stimuli, it is important to take into account real input signals to the nervous system. In the case of an insect with compound eyes, sceneries projected onto compound eye are the basic signals for visual information processing. In this paper, a method to reconstruct the projected images on the honeybee compound eyes is presented. This method estimates the illuminance change of a single ommatidium surface. It may be applied to analyze the correspondence between input images and retinal responses.

Introduction

Flying insects like bees can not only control their posture, but also navigate to the target based on time-varying image signals during their flight. Insect visual mechanisms are of interest for revealing neuronal principals of image processing and applying their simple structure to artificial vision. It is well known that capabilities of scenery recognition differ across species, so insects with compound eyes are living in a different visual world from mammals. However, only a few experiments take into consideration the difference in image perception by compound eyes [3]. It tends to be assumed that the same characteristics or information are used for recognition and behavioral control even in the insect brain. But if we want to reveal neural mechanisms for image processing in insect brains, it is essential to analyze and model based on real input signals into the photoreceptors.

In the case of honeybee flight, sceneries of her outside world are projected onto her compound eyes and received by photoreceptors in the ommatidium. Photoreceptor responses are processed through the visual signal pathway from retina to brain for detection of information and control of her behavior. Even though the projected images on her eyes are distorted by her position and posture, she can navigate easily with discrimination of patterns, recognition of shapes, and estimation of depth and distance. In my laboratory, we are building a mathematical model of image processing and flight control in the honeybee based on neural and behavioral properties. As a first step in the analysis of flight behavior to the visual stimulus, it is important to know real inputs to their eyes, that is, illuminance changes on the ommatidium surface. In this paper, a method to calculate the projection image on the honeybee compound eye is presented. A sequence of acquisition images through the flight path can be reconstructed by this method.

Section snippets

Model

A method to calculate the projection image on the compound eye of honeybees is presented and demonstrated by Giger [2]. In this method, a projection image of outside scenery was reconstructed by geometrical transformation, but I present a new method which is calculated based on illuminance changes on ommatidium lens surfaces.

Simulation results

Sceneries reconstructed by this illuminance change method were compared with the results of B-EYE [2]. It was confirmed that the illuminance change method also reconstructed a projection image on honeybee eyes. This method required a long time for calculations, because it estimated the illuminance for each ommatidia surface. However, it corresponded well to the photoreceptor model [1], since the presented method can produce real input stimuli for the photoreceptor.

An image compound of vertical,

Remarks

In this paper, a method was presented for reconstructing images projected onto the compound eye of worker honeybees. As shown in the simulation results, honeybees uses strongly distorted images whose distortion depends on the distance and direction from the target, but the honeybee shows skillful navigation based on this uncertain information. Presently, we are using a CCD camera to measure flight trajectories in the Y-shaped maze, which is used for conditioning honeybee [4]. Input images to

For further reading

The following reference may also be of interest to the reader [5].

Hidetoshi Ikeno received the B.E., M.E. and D.E. degree from the Toyohashi University of Technology in 1983, 1985 and 1993, respectively. 1985–1989: Research Associate; 1989–1992: Assistant Professor; 1994–1998: Associate Professor at Maizuru National College of Technology. Assistant Professor at Toyohashi University of Technology, 1992–1994. Since April 1998, he has been an Associate Professor at Himeji Institute of Technology.

References (6)

There are more references available in the full text version of this article.

Cited by (5)

Hidetoshi Ikeno received the B.E., M.E. and D.E. degree from the Toyohashi University of Technology in 1983, 1985 and 1993, respectively. 1985–1989: Research Associate; 1989–1992: Assistant Professor; 1994–1998: Associate Professor at Maizuru National College of Technology. Assistant Professor at Toyohashi University of Technology, 1992–1994. Since April 1998, he has been an Associate Professor at Himeji Institute of Technology.

A part of this study was performed through Special Coordination Fund (Neuroinformatics Research in Vision, PI Prof. Shiro Usui) for the promotion of Ministry of Education, Culture, Sports, Science and Technology of the Japanese Government.

View full text